"MST++: Multi-stage Spectral-wise Transformer for Efficient Spectral Reconstruction" (CVPRW 2022) & (Winner of NTIRE 2022 Challenge on Spectral Reconstruction from RGB)

Overview

MST++: Multi-stage Spectral-wise Transformer for Efficient Spectral Reconstruction (CVPRW 2022)

winner arXiv zhihu mst visitors

Yuanhao Cai, Jing Lin, Zudi Lin, Haoqian Wang, Yulun Zhang, Hanspeter Pfister, Radu Timofte, Luc Van Gool

The first two authors contribute equally to this work

News

  • 2022.04.17 : Our paper has been accepted by CVPRW 2022, code and models have been released. 🚀
  • 2022.04.02 : We win the First place of NTIRE 2022 Challenge on Spectral Reconstruction from RGB. 🏆
480 nm 520 nm 580 nm 660 nm

Abstract: Existing leading methods for spectral reconstruction (SR) focus on designing deeper or wider convolutional neural networks (CNNs) to learn the end-to-end mapping from the RGB image to its hyperspectral image (HSI). These CNN-based methods achieve impressive restoration performance while showing limitations in capturing the long-range dependencies and self-similarity prior. To cope with this problem, we propose a novel Transformer-based method, Multi-stage Spectral-wise Transformer (MST++), for efficient spectral reconstruction. In particular, we employ Spectral-wise Multi-head Self-attention (S-MSA) that is based on the HSI spatially sparse while spectrally self-similar nature to compose the basic unit, Spectral-wise Attention Block (SAB). Then SABs build up Single-stage Spectral-wise Transformer (SST) that exploits a U-shaped structure to extract multi-resolution contextual information. Finally, our MST++, cascaded by several SSTs, progressively improves the reconstruction quality from coarse to fine. Comprehensive experiments show that our MST++ significantly outperforms other state-of-the-art methods. In the NTIRE 2022 Spectral Reconstruction Challenge, our approach won the First place.


Network Architecture

Illustration of MST

Our MST++ is mainly based on our work MST, which is accepted by CVPR 2022.

Comparison with State-of-the-art Methods

This repo is a baseline and toolbox containing 11 image restoration algorithms for Spectral Reconstruction.

We are going to enlarge our model zoo in the future.

Supported algorithms:

comparison_fig

Results on NTIRE 2022 HSI Dataset - Validation

Method Params (M) FLOPS (G) MRAE RMSE PSNR Model Zoo
HSCNN+ 4.65 304.45 0.3814 0.0588 26.36 Google Drive / Baidu Disk
HRNet 31.70 163.81 0.3476 0.0550 26.89 Google Drive / Baidu Disk
EDSR 2.42 158.32 0.3277 0.0437 28.29 Google Drive / Baidu Disk
AWAN 4.04 270.61 0.2500 0.0367 31.22 Google Drive / Baidu Disk
HDNet 2.66 173.81 0.2048 0.0317 32.13 Google Drive / Baidu Disk
HINet 5.21 31.04 0.2032 0.0303 32.51 Google Drive / Baidu Disk
MIRNet 3.75 42.95 0.1890 0.0274 33.29 Google Drive / Baidu Disk
Restormer 15.11 93.77 0.1833 0.0274 33.40 Google Drive / Baidu Disk
MPRNet 3.62 101.59 0.1817 0.0270 33.50 Google Drive / Baidu Disk
MST-L 2.45 32.07 0.1772 0.0256 33.90 Google Drive / Baidu Disk
MST++ 1.62 23.05 0.1645 0.0248 34.32 Google Drive / Baidu Disk

Our MST++ siginificantly outperforms other methods while requiring cheaper Params and FLOPS.

Note: access code for Baidu Disk is mst1.

1. Create Envirement:

  • Python 3 (Recommend to use Anaconda)

  • NVIDIA GPU + CUDA

  • Python packages:

    cd MST-plus-plus
    pip install -r requirements.txt

2. Data Preparation:

  • Download training spectral images (Google Drive / Baidu Disk, code: mst1), training RGB images (Google Drive / Baidu Disk), validation spectral images (Google Drive / Baidu Disk), validation RGB images (Google Drive / Baidu Disk), and testing RGB images (Google Drive / Baidu Disk) from the competition website of NTIRE 2022 Spectral Reconstruction Challenge.

  • Place the training spectral images and validation spectral images to /MST-plus-plus/dataset/Train_Spec/.

  • Place the training RGB images and validation RGB images to /MST-plus-plus/dataset/Train_RGB/.

  • Place the testing RGB images to /MST-plus-plus/dataset/Test_RGB/.

  • Then this repo is collected as the following form:

    |--MST-plus-plus
        |--test_challenge_code
        |--test_develop_code
        |--train_code  
        |--dataset 
            |--Train_Spec
                |--ARAD_1K_0001.mat
                |--ARAD_1K_0002.mat
                : 
                |--ARAD_1K_0950.mat
      	|--Train_RGB
                |--ARAD_1K_0001.jpg
                |--ARAD_1K_0002.jpg
                : 
                |--ARAD_1K_0950.jpg
            |--Test_RGB
                |--ARAD_1K_0951.jpg
                |--ARAD_1K_0952.jpg
                : 
                |--ARAD_1K_1000.jpg
            |--split_txt
                |--train_list.txt
                |--valid_list.txt

3. Evaluation on the Validation Set:

(1) Download the pretrained model zoo from (Google Drive / Baidu Disk, code: mst1) and place them to /MST-plus-plus/test_develop_code/model_zoo/.

(2) Run the following command to test the model on the validation RGB images.

cd /MST-plus-plus/test_develop_code/

# test MST++
python test.py --data_root ../dataset/  --method mst_plus_plus --pretrained_model_path ./model_zoo/mst_plus_plus.pth --outf ./exp/mst_plus_plus/  --gpu_id 0

# test MST-L
python test.py --data_root ../dataset/  --method mst --pretrained_model_path ./model_zoo/mst.pth --outf ./exp/mst/  --gpu_id 0

# test MIRNet
python test.py --data_root ../dataset/  --method mirnet --pretrained_model_path ./model_zoo/mirnet.pth --outf ./exp/mirnet/  --gpu_id 0

# test HINet
python test.py --data_root ../dataset/  --method hinet --pretrained_model_path ./model_zoo/hinet.pth --outf ./exp/hinet/  --gpu_id 0

# test MPRNet
python test.py --data_root ../dataset/  --method mprnet --pretrained_model_path ./model_zoo/mprnet.pth --outf ./exp/mprnet/  --gpu_id 0

# test Restormer
python test.py --data_root ../dataset/  --method restormer --pretrained_model_path ./model_zoo/restormer.pth --outf ./exp/restormer/  --gpu_id 0

# test EDSR
python test.py --data_root ../dataset/  --method edsr --pretrained_model_path ./model_zoo/edsr.pth --outf ./exp/edsr/  --gpu_id 0

# test HDNet
python test.py --data_root ../dataset/  --method hdnet --pretrained_model_path ./model_zoo/hdnet.pth --outf ./exp/hdnet/  --gpu_id 0

# test HRNet
python test.py --data_root ../dataset/  --method hrnet --pretrained_model_path ./model_zoo/hrnet.pth --outf ./exp/hrnet/  --gpu_id 0

# test HSCNN+
python test.py --data_root ../dataset/  --method hscnn_plus --pretrained_model_path ./model_zoo/hscnn_plus.pth --outf ./exp/hscnn_plus/  --gpu_id 0

# test AWAN
python test.py --data_root ../dataset/  --method awan --pretrained_model_path ./model_zoo/awan.pth --outf ./exp/awan/  --gpu_id 0

The results will be saved in /MST-plus-plus/test_develop_code/exp/ in the mat format and the evaluation metric (including MRAE,RMSE,PSNR) will be printed.

4. Evaluation on the Test Set:

(1) Download the pretrained model zoo from (Google Drive / Baidu Disk, code: mst1) and place them to /MST-plus-plus/test_challenge_code/model_zoo/.

(2) Run the following command to test the model on the testing RGB images.

cd /MST-plus-plus/test_challenge_code/

# test MST++
python test.py --data_root ../dataset/  --method mst_plus_plus --pretrained_model_path ./model_zoo/mst_plus_plus.pth --outf ./exp/mst_plus_plus/  --gpu_id 0

# test MST-L
python test.py --data_root ../dataset/  --method mst --pretrained_model_path ./model_zoo/mst.pth --outf ./exp/mst/  --gpu_id 0

# test MIRNet
python test.py --data_root ../dataset/  --method mirnet --pretrained_model_path ./model_zoo/mirnet.pth --outf ./exp/mirnet/  --gpu_id 0

# test HINet
python test.py --data_root ../dataset/  --method hinet --pretrained_model_path ./model_zoo/hinet.pth --outf ./exp/hinet/  --gpu_id 0

# test MPRNet
python test.py --data_root ../dataset/  --method mprnet --pretrained_model_path ./model_zoo/mprnet.pth --outf ./exp/mprnet/  --gpu_id 0

# test Restormer
python test.py --data_root ../dataset/  --method restormer --pretrained_model_path ./model_zoo/restormer.pth --outf ./exp/restormer/  --gpu_id 0

# test EDSR
python test.py --data_root ../dataset/  --method edsr --pretrained_model_path ./model_zoo/edsr.pth --outf ./exp/edsr/  --gpu_id 0

# test HDNet
python test.py --data_root ../dataset/  --method hdnet --pretrained_model_path ./model_zoo/hdnet.pth --outf ./exp/hdnet/  --gpu_id 0

# test HRNet
python test.py --data_root ../dataset/  --method hrnet --pretrained_model_path ./model_zoo/hrnet.pth --outf ./exp/hrnet/  --gpu_id 0

# test HSCNN+
python test.py --data_root ../dataset/  --method hscnn_plus --pretrained_model_path ./model_zoo/hscnn_plus.pth --outf ./exp/hscnn_plus/  --gpu_id 0

The results and submission.zip will be saved in /MST-plus-plus/test_challenge_code/exp/.

5. Training

To train a model, run

cd /MST-plus-plus/train_code/

# train MST++
python train.py --method mst_plus_plus  --batch_size 20 --end_epoch 300 --init_lr 4e-4 --outf ./exp/mst_plus_plus/ --data_root ../dataset/  --patch_size 128 --stride 8  --gpu_id 0

# train MST-L
python train.py --method mst  --batch_size 20 --end_epoch 300 --init_lr 4e-4 --outf ./exp/mst/ --data_root ../dataset/  --patch_size 128 --stride 8  --gpu_id 0

# train MIRNet
python train.py --method mirnet  --batch_size 20 --end_epoch 300 --init_lr 4e-4 --outf ./exp/mirnet/ --data_root ../dataset/  --patch_size 128 --stride 8  --gpu_id 0

# train HINet
python train.py --method hinet  --batch_size 20 --end_epoch 300 --init_lr 2e-4 --outf ./exp/hinet/ --data_root ../dataset/  --patch_size 128 --stride 8  --gpu_id 0

# train MPRNet
python train.py --method mprnet  --batch_size 20 --end_epoch 300 --init_lr 2e-4 --outf ./exp/mprnet/ --data_root ../dataset/  --patch_size 128 --stride 8  --gpu_id 0

# train Restormer
python train.py --method restormer  --batch_size 20 --end_epoch 300 --init_lr 2e-4 --outf ./exp/restormer/ --data_root ../dataset/  --patch_size 128 --stride 8  --gpu_id 0

# train EDSR
python train.py --method edsr  --batch_size 20 --end_epoch 300 --init_lr 1e-4 --outf ./exp/edsr/ --data_root ../dataset/  --patch_size 128 --stride 8  --gpu_id 0

# train HDNet
python train.py --method hdnet  --batch_size 20 --end_epoch 300 --init_lr 4e-4 --outf ./exp/hdnet/ --data_root ../dataset/  --patch_size 128 --stride 8  --gpu_id 0

# train HRNet
python train.py --method hrnet  --batch_size 20 --end_epoch 300 --init_lr 1e-4 --outf ./exp/hrnet/ --data_root ../dataset/  --patch_size 128 --stride 8  --gpu_id 0

# train HSCNN+
python train.py --method hscnn_plus  --batch_size 20 --end_epoch 300 --init_lr 2e-4 --outf ./exp/hscnn_plus/ --data_root ../dataset/  --patch_size 128 --stride 8  --gpu_id 0

# train AWAN
python train.py --method awan  --batch_size 20 --end_epoch 300 --init_lr 1e-4 --outf ./exp/awan/ --data_root ../dataset/  --patch_size 128 --stride 8  --gpu_id 0

The training log and models will be saved in /MST-plus-plus/train_code/exp/.

6. Prediction:

(1) Download the pretrained model zoo from (Google Drive / Baidu Disk, code: mst1) and place them to /MST-plus-plus/predict_code/model_zoo/.

(2) Run the following command to reconstruct your own RGB image.

cd /MST-plus-plus/predict_code/

# reconstruct by MST++
python test.py --rgb_path ./demo/ARAD_1K_0912.jpg  --method mst_plus_plus --pretrained_model_path ./model_zoo/mst_plus_plus.pth --outf ./exp/mst_plus_plus/  --gpu_id 0

# reconstruct by MST-L
python test.py --rgb_path ./demo/ARAD_1K_0912.jpg  --method mst --pretrained_model_path ./model_zoo/mst.pth --outf ./exp/mst/  --gpu_id 0

# reconstruct by MIRNet
python test.py --rgb_path ./demo/ARAD_1K_0912.jpg  --method mirnet --pretrained_model_path ./model_zoo/mirnet.pth --outf ./exp/mirnet/  --gpu_id 0

# reconstruct by HINet
python test.py --rgb_path ./demo/ARAD_1K_0912.jpg  --method hinet --pretrained_model_path ./model_zoo/hinet.pth --outf ./exp/hinet/  --gpu_id 0

# reconstruct by MPRNet
python test.py --rgb_path ./demo/ARAD_1K_0912.jpg  --method mprnet --pretrained_model_path ./model_zoo/mprnet.pth --outf ./exp/mprnet/  --gpu_id 0

# reconstruct by Restormer
python test.py --rgb_path ./demo/ARAD_1K_0912.jpg  --method restormer --pretrained_model_path ./model_zoo/restormer.pth --outf ./exp/restormer/  --gpu_id 0

# reconstruct by EDSR
python test.py --rgb_path ./demo/ARAD_1K_0912.jpg --method edsr --pretrained_model_path ./model_zoo/edsr.pth --outf ./exp/edsr/  --gpu_id 0

# reconstruct by HDNet
python test.py --rgb_path ./demo/ARAD_1K_0912.jpg  --method hdnet --pretrained_model_path ./model_zoo/hdnet.pth --outf ./exp/hdnet/  --gpu_id 0

# reconstruct by HRNet
python test.py --rgb_path ./demo/ARAD_1K_0912.jpg  --method hrnet --pretrained_model_path ./model_zoo/hrnet.pth --outf ./exp/hrnet/  --gpu_id 0

# reconstruct by HSCNN+
python test.py --rgb_path ./demo/ARAD_1K_0912.jpg  --method hscnn_plus --pretrained_model_path ./model_zoo/hscnn_plus.pth --outf ./exp/hscnn_plus/  --gpu_id 0

You can replace './demo/ARAD_1K_0912.jpg' with your RGB image path. The reconstructed results will be saved in /MST-plus-plus/predict_code/exp/.

Citation

If this repo helps you, please consider citing our works:

@inproceedings{mst,
	title={Mask-guided Spectral-wise Transformer for Efficient Hyperspectral Image Reconstruction},
	author={Yuanhao Cai and Jing Lin and Xiaowan Hu and Haoqian Wang and Xin Yuan and Yulun Zhang and Radu Timofte and Luc Van Gool},
	booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
	year={2022}
}

@inproceedings{mst_pp,
  title={MST++: Multi-stage Spectral-wise Transformer for Efficient Spectral Reconstruction},
  author={Yuanhao Cai and Jing Lin and Zudi Lin and Haoqian Wang and Yulun Zhang and Hanspeter Pfister and Radu Timofte and Luc Van Gool},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops},
  year={2022}
}

@inproceedings{hdnet,
	title={HDNet: High-resolution Dual-domain Learning for Spectral Compressive Imaging},
	author={Xiaowan Hu and Yuanhao Cai and Jing Lin and  Haoqian Wang and Xin Yuan and Yulun Zhang and Radu Timofte and Luc Van Gool},
	booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
	year={2022}
}
Comments
  • How to get the predict results

    How to get the predict results

    Thanks for your open-source code. The MST++ is an amzing project in HSI reconstruction scene. But your code only have train and test code, which not contains the predict code. I'm a people of a new type of HSI reconstruction and I don't have any idea about the predict results. So, May you open your predict code in your repository?

    opened by randomNNN 16
  • MST++的PSNR34.32是ensembel 以后的结果吗?

    MST++的PSNR34.32是ensembel 以后的结果吗?

    作者您好,您的工作非常棒,令人激动!我有三个疑问,希望您能解答: 1)MST++的PSNR34.32是ensembel 以后的结果吗 2)train了300个epoch以后,是使用最后的pth来进行计算指标吗,还是根据测试集上的指标来选一个最好的模型 3)我按照您设置的参数进行训练,不知为何train了395个epoch... 训练命令(一张3090显卡):python train.py --method mst_plus_plus --batch_size 20 --end_epoch 300 --init_lr 4e-4 --outf ./exp/mst_plus_plus/ --patch_size 128 --stride 8 --gpu_id 0 [iter:395460/300000],lr=0.000092650,train_losses.avg=0.102350816 [iter:395480/300000],lr=0.000092685,train_losses.avg=0.102352098 [iter:395500/300000],lr=0.000092720,train_losses.avg=0.102352992 [iter:395520/300000],lr=0.000092755,train_losses.avg=0.102354914 [iter:395540/300000],lr=0.000092791,train_losses.avg=0.102356508 [iter:395560/300000],lr=0.000092826,train_losses.avg=0.102357790 然后,我用最后的模型跑出来的指标为: load model from ../train_code/exp/mst_plus_plus/2022_04_23_14_46_46/net_395epoch.pth method:mst_plus_plus, mrae:0.2037583440542221, rmse:0.029374651610851288, psnr:32.78853988647461 我感觉和您文中的指标还有一定差距,您觉得这正常吗,或者是有什么原因造成的(有别的epoch能跑到33.6左右的指标) 希望能得到您的解答!

    opened by chenjiachengzzz 11
  • HDNet code

    HDNet code

    It seems that the spatial_attention and spectral_attention(SDL_attention) code implementation of the HDNet is contrary to what you written in the paper

    opened by lzw-lzw 10
  • MST 训练

    MST 训练

    尊敬的作者您好, python train.py --method mst --batch_size 20 --end_epoch 300 --init_lr 4e-4 --outf ./exp/mst/ --data_root ../dataset/ --patch_size 128 --stride 8 --gpu_id 0 使用上面的命令运行,得到的结果是这样的,别的模型是可以正常跑通的,只有这个mst,是我哪里弄错了吗?希望能得到您的解答! 2022-04-23 09:47:00 - Iter[001000], Epoch[000001], learning rate : 0.000399989, Train Loss: 0.646447778, Test MRAE: 0.531796575, Test RMSE: 0.091432519, Test PSNR: 19.090953827 2022-04-23 10:02:48 - Iter[002000], Epoch[000002], learning rate : 0.000399956, Train Loss: 0.578481972, Test MRAE: 0.491525143, Test RMSE: 0.086475834, Test PSNR: 19.247301102 2022-04-23 10:18:36 - Iter[003000], Epoch[000003], learning rate : 0.000399902, Train Loss: 0.547274351, Test MRAE: 0.494777530, Test RMSE: 0.079640247, Test PSNR: 19.196660995 2022-04-23 10:34:24 - Iter[004000], Epoch[000004], learning rate : 0.000399825, Train Loss: 0.526349247, Test MRAE: 0.472677380, Test RMSE: 0.069328219, Test PSNR: 19.027240753 2022-04-23 10:50:12 - Iter[005000], Epoch[000005], learning rate : 0.000399727, Train Loss: 0.511086941, Test MRAE: 0.358050972, Test RMSE: 0.059412364, Test PSNR: 19.215837479 2022-04-23 11:06:00 - Iter[006000], Epoch[000006], learning rate : 0.000399606, Train Loss: 0.499529332, Test MRAE: 0.356865495, Test RMSE: 0.055258289, Test PSNR: 19.074251175 2022-04-23 11:21:49 - Iter[007000], Epoch[000007], learning rate : 0.000399464, Train Loss: 0.489558429, Test MRAE: 0.378400117, Test RMSE: 0.057141136, Test PSNR: 19.100608826 2022-04-23 11:37:37 - Iter[008000], Epoch[000008], learning rate : 0.000399301, Train Loss: 0.481531292, Test MRAE: 0.362446398, Test RMSE: 0.056064066, Test PSNR: 19.177246094 2022-04-23 11:53:25 - Iter[009000], Epoch[000009], learning rate : 0.000399115, Train Loss: 0.474746048, Test MRAE: 0.331809163, Test RMSE: 0.051642809, Test PSNR: 19.225530624 2022-04-23 12:09:13 - Iter[010000], Epoch[000010], learning rate : 0.000398907, Train Loss: 0.468800724, Test MRAE: 0.300195664, Test RMSE: 0.046801656, Test PSNR: 19.080495834 2022-04-23 12:25:01 - Iter[011000], Epoch[000011], learning rate : 0.000398678, Train Loss: 0.463295877, Test MRAE: 0.327608109, Test RMSE: 0.050521433, Test PSNR: 19.182096481 2022-04-23 12:40:49 - Iter[012000], Epoch[000012], learning rate : 0.000398427, Train Loss: 0.458517045, Test MRAE: 0.422704875, Test RMSE: 0.066085078, Test PSNR: 19.202255249 2022-04-23 12:56:38 - Iter[013000], Epoch[000013], learning rate : 0.000398154, Train Loss: 0.453951567, Test MRAE: 0.444770008, Test RMSE: 0.066878960, Test PSNR: 18.973522186 2022-04-23 13:12:27 - Iter[014000], Epoch[000014], learning rate : 0.000397860, Train Loss: 0.447406828, Test MRAE: 0.336395442, Test RMSE: 0.050711267, Test PSNR: 19.189912796 2022-04-23 13:28:16 - Iter[015000], Epoch[000015], learning rate : 0.000397544, Train Loss: 0.440287501, Test MRAE: 0.364379525, Test RMSE: 0.054785427, Test PSNR: 19.161180496 2022-04-23 13:44:06 - Iter[016000], Epoch[000016], learning rate : 0.000397207, Train Loss: 0.432897270, Test MRAE: 0.370432645, Test RMSE: 0.056947071, Test PSNR: 19.268909454 2022-04-23 13:59:54 - Iter[017000], Epoch[000017], learning rate : 0.000396847, Train Loss: 0.425944477, Test MRAE: 0.386497170, Test RMSE: 0.057578120, Test PSNR: 19.306638718 2022-04-23 14:15:41 - Iter[018000], Epoch[000018], learning rate : 0.000396467, Train Loss: 0.419327229, Test MRAE: 0.381819218, Test RMSE: 0.054971345, Test PSNR: 19.137437820 2022-04-23 14:31:30 - Iter[019000], Epoch[000019], learning rate : 0.000396065, Train Loss: 0.412961543, Test MRAE: 0.268086284, Test RMSE: 0.039695993, Test PSNR: 19.237449646 2022-04-23 14:47:18 - Iter[020000], Epoch[000020], learning rate : 0.000395641, Train Loss: 0.406925291, Test MRAE: 0.278270304, Test RMSE: 0.040124148, Test PSNR: 19.128618240 2022-04-23 15:03:06 - Iter[021000], Epoch[000021], learning rate : 0.000395196, Train Loss: 0.401020914, Test MRAE: 0.263858318, Test RMSE: 0.039533190, Test PSNR: 19.288784027 2022-04-23 15:18:55 - Iter[022000], Epoch[000022], learning rate : 0.000394729, Train Loss: 0.395406336, Test MRAE: 0.293986678, Test RMSE: 0.042647284, Test PSNR: 19.238164902 2022-04-23 15:34:43 - Iter[023000], Epoch[000023], learning rate : 0.000394242, Train Loss: 0.390212119, Test MRAE: 0.291280866, Test RMSE: 0.042363208, Test PSNR: 19.284023285 2022-04-23 15:50:32 - Iter[024000], Epoch[000024], learning rate : 0.000393733, Train Loss: 0.385213077, Test MRAE: 0.233923718, Test RMSE: 0.036982559, Test PSNR: 19.216468811 2022-04-23 16:06:20 - Iter[025000], Epoch[000025], learning rate : 0.000393203, Train Loss: 0.380559415, Test MRAE: 0.260465741, Test RMSE: 0.039078295, Test PSNR: 19.235170364 2022-04-23 16:22:09 - Iter[026000], Epoch[000026], learning rate : 0.000392651, Train Loss: 0.375919461, Test MRAE: 0.254196465, Test RMSE: 0.036160678, Test PSNR: 19.114507675 2022-04-23 16:37:58 - Iter[027000], Epoch[000027], learning rate : 0.000392079, Train Loss: 0.371591926, Test MRAE: 0.314058006, Test RMSE: 0.044734038, Test PSNR: 19.360179901 2022-04-23 16:53:49 - Iter[028000], Epoch[000028], learning rate : 0.000391486, Train Loss: 0.367457062, Test MRAE: 0.226818457, Test RMSE: 0.031798869, Test PSNR: 18.716556549 2022-04-23 17:09:37 - Iter[029000], Epoch[000029], learning rate : 0.000390872, Train Loss: 0.363370985, Test MRAE: 0.271478027, Test RMSE: 0.041398086, Test PSNR: 19.368923187 2022-04-23 17:25:25 - Iter[030000], Epoch[000030], learning rate : 0.000390236, Train Loss: 0.359518796, Test MRAE: 0.293181419, Test RMSE: 0.040424421, Test PSNR: 19.312316895 2022-04-23 17:41:13 - Iter[031000], Epoch[000031], learning rate : 0.000389580, Train Loss: 0.355821848, Test MRAE: 0.245824501, Test RMSE: 0.034910016, Test PSNR: 19.084083557 2022-04-23 17:57:02 - Iter[032000], Epoch[000032], learning rate : 0.000388904, Train Loss: 0.352144718, Test MRAE: 0.266790211, Test RMSE: 0.040394772, Test PSNR: 19.260843277 2022-04-23 18:12:53 - Iter[033000], Epoch[000033], learning rate : 0.000388206, Train Loss: 0.348656774, Test MRAE: 0.222802311, Test RMSE: 0.034814272, Test PSNR: 19.181829453 2022-04-23 18:28:42 - Iter[034000], Epoch[000034], learning rate : 0.000387488, Train Loss: 0.345295727, Test MRAE: 0.298925221, Test RMSE: 0.043574888, Test PSNR: 19.372837067 2022-04-23 18:44:32 - Iter[035000], Epoch[000035], learning rate : 0.000386750, Train Loss: 0.342086405, Test MRAE: 0.234956443, Test RMSE: 0.036083721, Test PSNR: 19.267917633 2022-04-23 19:00:22 - Iter[036000], Epoch[000036], learning rate : 0.000385991, Train Loss: 0.338931412, Test MRAE: 0.228018716, Test RMSE: 0.032620184, Test PSNR: 19.047657013 2022-04-23 19:16:11 - Iter[037000], Epoch[000037], learning rate : 0.000385212, Train Loss: 0.335954189, Test MRAE: 0.229232669, Test RMSE: 0.034972440, Test PSNR: 19.219429016 2022-04-23 19:31:59 - Iter[038000], Epoch[000038], learning rate : 0.000384413, Train Loss: 0.333125830, Test MRAE: 0.230264142, Test RMSE: 0.034406129, Test PSNR: 19.242214203 2022-04-23 19:47:48 - Iter[039000], Epoch[000039], learning rate : 0.000383593, Train Loss: 0.330276132, Test MRAE: 0.274099082, Test RMSE: 0.043192532, Test PSNR: 19.348083496 2022-04-23 20:03:35 - Iter[040000], Epoch[000040], learning rate : 0.000382753, Train Loss: 0.327579051, Test MRAE: 0.218261853, Test RMSE: 0.031082967, Test PSNR: 18.932905197 2022-04-23 20:19:24 - Iter[041000], Epoch[000041], learning rate : 0.000381893, Train Loss: 0.324890018, Test MRAE: 0.254400879, Test RMSE: 0.038573589, Test PSNR: 19.314989090 2022-04-23 20:35:13 - Iter[042000], Epoch[000042], learning rate : 0.000381014, Train Loss: 0.322247148, Test MRAE: 0.259322703, Test RMSE: 0.038424168, Test PSNR: 19.213300705 2022-04-23 20:51:02 - Iter[043000], Epoch[000043], learning rate : 0.000380115, Train Loss: 0.319739074, Test MRAE: 0.238067225, Test RMSE: 0.035432223, Test PSNR: 19.308258057 2022-04-23 21:06:51 - Iter[044000], Epoch[000044], learning rate : 0.000379195, Train Loss: 0.317296684, Test MRAE: 0.216048419, Test RMSE: 0.031597815, Test PSNR: 19.090858459 2022-04-23 21:22:38 - Iter[045000], Epoch[000045], learning rate : 0.000378257, Train Loss: 0.314926445, Test MRAE: 0.235811070, Test RMSE: 0.034594744, Test PSNR: 19.259517670 2022-04-23 21:38:27 - Iter[046000], Epoch[000046], learning rate : 0.000377299, Train Loss: 0.312630028, Test MRAE: 0.209554464, Test RMSE: 0.030831426, Test PSNR: 19.153333664 2022-04-23 21:54:16 - Iter[047000], Epoch[000047], learning rate : 0.000376321, Train Loss: 0.310319424, Test MRAE: 0.214102373, Test RMSE: 0.031406451, Test PSNR: 19.192979813 2022-04-23 22:10:04 - Iter[048000], Epoch[000048], learning rate : 0.000375324, Train Loss: 0.308094501, Test MRAE: 0.199715927, Test RMSE: 0.030652732, Test PSNR: 19.189508438 2022-04-23 22:25:51 - Iter[049000], Epoch[000049], learning rate : 0.000374308, Train Loss: 0.305966139, Test MRAE: 0.218578279, Test RMSE: 0.031519547, Test PSNR: 19.050167084 2022-04-23 22:41:40 - Iter[050000], Epoch[000050], learning rate : 0.000373273, Train Loss: 0.303862125, Test MRAE: 0.215297714, Test RMSE: 0.032994971, Test PSNR: 19.179225922 2022-04-23 22:57:28 - Iter[051000], Epoch[000051], learning rate : 0.000372219, Train Loss: 0.301793545, Test MRAE: 0.231257230, Test RMSE: 0.032595847, Test PSNR: 19.175262451 2022-04-23 23:13:17 - Iter[052000], Epoch[000052], learning rate : 0.000371146, Train Loss: 0.299778074, Test MRAE: 0.253783792, Test RMSE: 0.037507851, Test PSNR: 19.363800049 2022-04-23 23:29:05 - Iter[053000], Epoch[000053], learning rate : 0.000370055, Train Loss: 0.297876358, Test MRAE: 0.258977175, Test RMSE: 0.040128287, Test PSNR: 19.314584732 2022-04-23 23:44:53 - Iter[054000], Epoch[000054], learning rate : 0.000368945, Train Loss: 0.295973837, Test MRAE: 0.217594683, Test RMSE: 0.032133736, Test PSNR: 19.153348923 2022-04-24 00:00:41 - Iter[055000], Epoch[000055], learning rate : 0.000367816, Train Loss: 0.294095606, Test MRAE: 0.216711819, Test RMSE: 0.031819437, Test PSNR: 19.180938721 2022-04-24 00:16:29 - Iter[056000], Epoch[000056], learning rate : 0.000366669, Train Loss: 0.292232960, Test MRAE: 0.207258597, Test RMSE: 0.029869573, Test PSNR: 19.150335312 2022-04-24 00:32:17 - Iter[057000], Epoch[000057], learning rate : 0.000365504, Train Loss: 0.290452927, Test MRAE: 0.198026657, Test RMSE: 0.028679363, Test PSNR: 19.005662918 2022-04-24 00:48:05 - Iter[058000], Epoch[000058], learning rate : 0.000364320, Train Loss: 0.288685858, Test MRAE: 0.221201345, Test RMSE: 0.033238016, Test PSNR: 19.217224121 2022-04-24 01:03:54 - Iter[059000], Epoch[000059], learning rate : 0.000363119, Train Loss: 0.287011743, Test MRAE: 0.213097245, Test RMSE: 0.032143150, Test PSNR: 19.209981918 2022-04-24 01:19:42 - Iter[060000], Epoch[000060], learning rate : 0.000361900, Train Loss: 0.285367638, Test MRAE: 0.193839341, Test RMSE: 0.027465345, Test PSNR: 18.971504211 2022-04-24 01:35:30 - Iter[061000], Epoch[000061], learning rate : 0.000360663, Train Loss: 0.283741534, Test MRAE: 0.196817130, Test RMSE: 0.029001579, Test PSNR: 19.062067032 2022-04-24 01:51:19 - Iter[062000], Epoch[000062], learning rate : 0.000359409, Train Loss: 0.282072276, Test MRAE: 0.219744995, Test RMSE: 0.029378273, Test PSNR: 18.866895676 2022-04-24 02:07:07 - Iter[063000], Epoch[000063], learning rate : 0.000358137, Train Loss: 0.280501366, Test MRAE: 0.229305848, Test RMSE: 0.033509906, Test PSNR: 19.214229584 2022-04-24 02:22:55 - Iter[064000], Epoch[000064], learning rate : 0.000356848, Train Loss: 0.278931111, Test MRAE: 0.206814021, Test RMSE: 0.030476322, Test PSNR: 19.040904999 2022-04-24 02:38:42 - Iter[065000], Epoch[000065], learning rate : 0.000355542, Train Loss: 0.277395368, Test MRAE: 0.208180115, Test RMSE: 0.031854365, Test PSNR: 19.156114578 2022-04-24 02:54:30 - Iter[066000], Epoch[000066], learning rate : 0.000354219, Train Loss: 0.275906801, Test MRAE: 0.195947483, Test RMSE: 0.030340478, Test PSNR: 19.152935028 2022-04-24 03:10:18 - Iter[067000], Epoch[000067], learning rate : 0.000352879, Train Loss: 0.274478436, Test MRAE: 0.220566273, Test RMSE: 0.032208432, Test PSNR: 18.987400055 2022-04-24 03:26:06 - Iter[068000], Epoch[000068], learning rate : 0.000351522, Train Loss: 0.273031890, Test MRAE: 0.198420197, Test RMSE: 0.029046385, Test PSNR: 18.940135956 2022-04-24 03:41:54 - Iter[069000], Epoch[000069], learning rate : 0.000350149, Train Loss: 0.271648794, Test MRAE: 0.240019321, Test RMSE: 0.034521896, Test PSNR: 19.117261887 2022-04-24 03:57:42 - Iter[070000], Epoch[000070], learning rate : 0.000348759, Train Loss: 0.270240217, Test MRAE: 0.203085661, Test RMSE: 0.028956201, Test PSNR: 19.012472153 2022-04-24 04:13:31 - Iter[071000], Epoch[000071], learning rate : 0.000347353, Train Loss: 0.268938541, Test MRAE: 0.213608027, Test RMSE: 0.031390432, Test PSNR: 18.951906204 2022-04-24 04:29:19 - Iter[072000], Epoch[000072], learning rate : 0.000345931, Train Loss: 0.267599642, Test MRAE: 0.231374159, Test RMSE: 0.031735256, Test PSNR: 19.098361969 2022-04-24 04:45:06 - Iter[073000], Epoch[000073], learning rate : 0.000344493, Train Loss: 0.266298503, Test MRAE: 0.250184625, Test RMSE: 0.035862882, Test PSNR: 19.221879959 2022-04-24 05:00:54 - Iter[074000], Epoch[000074], learning rate : 0.000343039, Train Loss: 0.264970332, Test MRAE: 0.216034293, Test RMSE: 0.032297533, Test PSNR: 19.119438171 2022-04-24 05:16:41 - Iter[075000], Epoch[000075], learning rate : 0.000341569, Train Loss: 0.263698131, Test MRAE: 0.236452579, Test RMSE: 0.034641251, Test PSNR: 19.267896652 2022-04-24 05:32:29 - Iter[076000], Epoch[000076], learning rate : 0.000340084, Train Loss: 0.262492269, Test MRAE: 0.226200759, Test RMSE: 0.032904580, Test PSNR: 19.063945770 2022-04-24 05:48:17 - Iter[077000], Epoch[000077], learning rate : 0.000338584, Train Loss: 0.261239469, Test MRAE: 0.201683655, Test RMSE: 0.029232167, Test PSNR: 18.907796860 2022-04-24 06:04:04 - Iter[078000], Epoch[000078], learning rate : 0.000337069, Train Loss: 0.260039657, Test MRAE: 0.214427084, Test RMSE: 0.031692069, Test PSNR: 18.943357468 2022-04-24 06:19:52 - Iter[079000], Epoch[000079], learning rate : 0.000335538, Train Loss: 0.258839667, Test MRAE: 0.209781334, Test RMSE: 0.030609982, Test PSNR: 19.051986694 2022-04-24 06:35:40 - Iter[080000], Epoch[000080], learning rate : 0.000333993, Train Loss: 0.257642329, Test MRAE: 0.192540377, Test RMSE: 0.029445488, Test PSNR: 19.034675598 2022-04-24 06:51:25 - Iter[081000], Epoch[000081], learning rate : 0.000332433, Train Loss: 0.256517678, Test MRAE: 0.222092792, Test RMSE: 0.030401962, Test PSNR: 18.930913925 2022-04-24 07:07:08 - Iter[082000], Epoch[000082], learning rate : 0.000330859, Train Loss: 0.255394250, Test MRAE: 0.209056050, Test RMSE: 0.029050352, Test PSNR: 19.042034149 2022-04-24 07:22:51 - Iter[083000], Epoch[000083], learning rate : 0.000329270, Train Loss: 0.254281640, Test MRAE: 0.214589760, Test RMSE: 0.030067844, Test PSNR: 18.710792542 2022-04-24 07:38:34 - Iter[084000], Epoch[000084], learning rate : 0.000327668, Train Loss: 0.253181159, Test MRAE: 0.198045820, Test RMSE: 0.028749663, Test PSNR: 19.026756287 2022-04-24 07:54:19 - Iter[085000], Epoch[000085], learning rate : 0.000326051, Train Loss: 0.252079219, Test MRAE: 0.212200597, Test RMSE: 0.030908102, Test PSNR: 19.004436493 2022-04-24 08:10:03 - Iter[086000], Epoch[000086], learning rate : 0.000324421, Train Loss: 0.251013666, Test MRAE: 0.205034807, Test RMSE: 0.029491324, Test PSNR: 19.155452728 2022-04-24 08:25:51 - Iter[087000], Epoch[000087], learning rate : 0.000322777, Train Loss: 0.249932632, Test MRAE: 0.192795992, Test RMSE: 0.028440528, Test PSNR: 19.070756912 2022-04-24 08:41:33 - Iter[088000], Epoch[000088], learning rate : 0.000321119, Train Loss: 0.248863876, Test MRAE: 0.206967190, Test RMSE: 0.028357433, Test PSNR: 18.958072662 2022-04-24 08:57:17 - Iter[089000], Epoch[000089], learning rate : 0.000319449, Train Loss: 0.247781381, Test MRAE: 0.212967843, Test RMSE: 0.029828170, Test PSNR: 18.949298859 2022-04-24 09:12:59 - Iter[090000], Epoch[000090], learning rate : 0.000317765, Train Loss: 0.246814713, Test MRAE: 0.217871219, Test RMSE: 0.031028254, Test PSNR: 19.021213531 2022-04-24 09:28:41 - Iter[091000], Epoch[000091], learning rate : 0.000316068, Train Loss: 0.245769218, Test MRAE: 0.222220868, Test RMSE: 0.031629700, Test PSNR: 19.008470535 2022-04-24 09:44:25 - Iter[092000], Epoch[000092], learning rate : 0.000314359, Train Loss: 0.244765893, Test MRAE: 0.201765835, Test RMSE: 0.030243544, Test PSNR: 19.158527374 2022-04-24 10:00:07 - Iter[093000], Epoch[000093], learning rate : 0.000312637, Train Loss: 0.243786022, Test MRAE: 0.238795847, Test RMSE: 0.035489723, Test PSNR: 19.237535477 2022-04-24 10:15:52 - Iter[094000], Epoch[000094], learning rate : 0.000310903, Train Loss: 0.242815033, Test MRAE: 0.227278069, Test RMSE: 0.032577001, Test PSNR: 19.165273666 2022-04-24 10:31:36 - Iter[095000], Epoch[000095], learning rate : 0.000309157, Train Loss: 0.241861641, Test MRAE: 0.206680715, Test RMSE: 0.029225241, Test PSNR: 19.057720184 2022-04-24 10:47:19 - Iter[096000], Epoch[000096], learning rate : 0.000307399, Train Loss: 0.240916863, Test MRAE: 0.221876547, Test RMSE: 0.031988315, Test PSNR: 19.071466446 2022-04-24 11:03:01 - Iter[097000], Epoch[000097], learning rate : 0.000305629, Train Loss: 0.239972606, Test MRAE: 0.205407277, Test RMSE: 0.029455291, Test PSNR: 18.912952423 2022-04-24 11:18:43 - Iter[098000], Epoch[000098], learning rate : 0.000303848, Train Loss: 0.239050597, Test MRAE: 0.218140364, Test RMSE: 0.031495962, Test PSNR: 19.028364182 2022-04-24 11:34:26 - Iter[099000], Epoch[000099], learning rate : 0.000302056, Train Loss: 0.145563006, Test MRAE: 0.212117374, Test RMSE: 0.029292498, Test PSNR: 18.875003815 2022-04-24 11:50:08 - Iter[100000], Epoch[000100], learning rate : 0.000300252, Train Loss: 0.145217299, Test MRAE: 0.214958370, Test RMSE: 0.030232767, Test PSNR: 18.982784271 2022-04-24 12:05:50 - Iter[101000], Epoch[000101], learning rate : 0.000298437, Train Loss: 0.146453694, Test MRAE: 0.227019936, Test RMSE: 0.032480333, Test PSNR: 19.083282471 2022-04-24 12:21:38 - Iter[102000], Epoch[000102], learning rate : 0.000296612, Train Loss: 0.146322653, Test MRAE: 0.239566207, Test RMSE: 0.035496548, Test PSNR: 19.104885101 2022-04-24 12:37:20 - Iter[103000], Epoch[000103], learning rate : 0.000294776, Train Loss: 0.146285564, Test MRAE: 0.191646069, Test RMSE: 0.028251326, Test PSNR: 19.006746292 2022-04-24 12:53:03 - Iter[104000], Epoch[000104], learning rate : 0.000292929, Train Loss: 0.146255314, Test MRAE: 0.234831825, Test RMSE: 0.034441300, Test PSNR: 19.204271317 2022-04-24 13:08:45 - Iter[105000], Epoch[000105], learning rate : 0.000291073, Train Loss: 0.145497769, Test MRAE: 0.207711518, Test RMSE: 0.030052185, Test PSNR: 19.077249527 2022-04-24 13:24:27 - Iter[106000], Epoch[000106], learning rate : 0.000289207, Train Loss: 0.145523682, Test MRAE: 0.222439647, Test RMSE: 0.032102700, Test PSNR: 19.164697647 2022-04-24 13:40:12 - Iter[107000], Epoch[000107], learning rate : 0.000287330, Train Loss: 0.145412102, Test MRAE: 0.200026944, Test RMSE: 0.027787490, Test PSNR: 18.735790253 2022-04-24 13:56:02 - Iter[108000], Epoch[000108], learning rate : 0.000285445, Train Loss: 0.145686775, Test MRAE: 0.211092830, Test RMSE: 0.031725895, Test PSNR: 19.057603836 2022-04-24 14:11:52 - Iter[109000], Epoch[000109], learning rate : 0.000283550, Train Loss: 0.145138130, Test MRAE: 0.202462837, Test RMSE: 0.029094383, Test PSNR: 19.105033875 2022-04-24 14:27:39 - Iter[110000], Epoch[000110], learning rate : 0.000281646, Train Loss: 0.144788370, Test MRAE: 0.204958782, Test RMSE: 0.028470399, Test PSNR: 19.004680634 2022-04-24 14:43:25 - Iter[111000], Epoch[000111], learning rate : 0.000279733, Train Loss: 0.144440651, Test MRAE: 0.194457725, Test RMSE: 0.028255261, Test PSNR: 18.978731155 2022-04-24 14:59:13 - Iter[112000], Epoch[000112], learning rate : 0.000277811, Train Loss: 0.144009396, Test MRAE: 0.197285131, Test RMSE: 0.028200772, Test PSNR: 18.996498108 2022-04-24 15:15:00 - Iter[113000], Epoch[000113], learning rate : 0.000275881, Train Loss: 0.143576682, Test MRAE: 0.203258514, Test RMSE: 0.029270183, Test PSNR: 18.912971497 2022-04-24 15:30:45 - Iter[114000], Epoch[000114], learning rate : 0.000273943, Train Loss: 0.143229470, Test MRAE: 0.219165146, Test RMSE: 0.031687632, Test PSNR: 19.071453094 2022-04-24 15:46:33 - Iter[115000], Epoch[000115], learning rate : 0.000271996, Train Loss: 0.142740473, Test MRAE: 0.216698647, Test RMSE: 0.031449940, Test PSNR: 19.045347214 2022-04-24 16:02:23 - Iter[116000], Epoch[000116], learning rate : 0.000270042, Train Loss: 0.142507568, Test MRAE: 0.193366423, Test RMSE: 0.028129779, Test PSNR: 19.005863190 2022-04-24 16:18:10 - Iter[117000], Epoch[000117], learning rate : 0.000268080, Train Loss: 0.142188013, Test MRAE: 0.219309017, Test RMSE: 0.030824291, Test PSNR: 19.017368317 2022-04-24 16:33:56 - Iter[118000], Epoch[000118], learning rate : 0.000266111, Train Loss: 0.141825795, Test MRAE: 0.205177620, Test RMSE: 0.028586203, Test PSNR: 18.947324753 2022-04-24 16:49:44 - Iter[119000], Epoch[000119], learning rate : 0.000264134, Train Loss: 0.141454026, Test MRAE: 0.213753939, Test RMSE: 0.030987954, Test PSNR: 19.093702316 2022-04-24 17:05:34 - Iter[120000], Epoch[000120], learning rate : 0.000262151, Train Loss: 0.141122550, Test MRAE: 0.205288440, Test RMSE: 0.029451758, Test PSNR: 19.051704407 2022-04-24 17:21:21 - Iter[121000], Epoch[000121], learning rate : 0.000260161, Train Loss: 0.141675979, Test MRAE: 0.213665485, Test RMSE: 0.029897889, Test PSNR: 18.968690872 2022-04-24 17:37:06 - Iter[122000], Epoch[000122], learning rate : 0.000258164, Train Loss: 0.141259611, Test MRAE: 0.203018293, Test RMSE: 0.029807677, Test PSNR: 19.059270859 2022-04-24 17:52:52 - Iter[123000], Epoch[000123], learning rate : 0.000256161, Train Loss: 0.140892401, Test MRAE: 0.205029026, Test RMSE: 0.029058052, Test PSNR: 19.114631653 2022-04-24 18:08:38 - Iter[124000], Epoch[000124], learning rate : 0.000254152, Train Loss: 0.140450522, Test MRAE: 0.198189601, Test RMSE: 0.028176619, Test PSNR: 18.899404526 2022-04-24 18:24:25 - Iter[125000], Epoch[000125], learning rate : 0.000252136, Train Loss: 0.140061647, Test MRAE: 0.227593854, Test RMSE: 0.032393444, Test PSNR: 19.117650986 2022-04-24 18:40:12 - Iter[126000], Epoch[000126], learning rate : 0.000250116, Train Loss: 0.139804602, Test MRAE: 0.214596003, Test RMSE: 0.030325968, Test PSNR: 19.085477829 2022-04-24 18:56:03 - Iter[127000], Epoch[000127], learning rate : 0.000248089, Train Loss: 0.139489800, Test MRAE: 0.219026044, Test RMSE: 0.030452706, Test PSNR: 18.957645416 2022-04-24 19:11:50 - Iter[128000], Epoch[000128], learning rate : 0.000246058, Train Loss: 0.139124364, Test MRAE: 0.220908672, Test RMSE: 0.031795617, Test PSNR: 19.092975616 2022-04-24 19:27:36 - Iter[129000], Epoch[000129], learning rate : 0.000244022, Train Loss: 0.138711065, Test MRAE: 0.223648518, Test RMSE: 0.030946231, Test PSNR: 19.023336411 2022-04-24 19:43:24 - Iter[130000], Epoch[000130], learning rate : 0.000241980, Train Loss: 0.138381705, Test MRAE: 0.199091196, Test RMSE: 0.028440719, Test PSNR: 18.978559494 2022-04-24 19:59:10 - Iter[131000], Epoch[000131], learning rate : 0.000239935, Train Loss: 0.138126105, Test MRAE: 0.200671941, Test RMSE: 0.028158125, Test PSNR: 18.979644775 2022-04-24 20:14:55 - Iter[132000], Epoch[000132], learning rate : 0.000237885, Train Loss: 0.137827173, Test MRAE: 0.203352720, Test RMSE: 0.028771115, Test PSNR: 18.875143051 2022-04-24 20:30:43 - Iter[133000], Epoch[000133], learning rate : 0.000235830, Train Loss: 0.137512103, Test MRAE: 0.251916736, Test RMSE: 0.035541732, Test PSNR: 19.041845322 2022-04-24 20:46:31 - Iter[134000], Epoch[000134], learning rate : 0.000233772, Train Loss: 0.137238741, Test MRAE: 0.229384869, Test RMSE: 0.032080282, Test PSNR: 18.928392410 2022-04-24 21:02:16 - Iter[135000], Epoch[000135], learning rate : 0.000231711, Train Loss: 0.136950210, Test MRAE: 0.230571747, Test RMSE: 0.033738092, Test PSNR: 19.063541412 2022-04-24 21:18:02 - Iter[136000], Epoch[000136], learning rate : 0.000229646, Train Loss: 0.136655480, Test MRAE: 0.210024893, Test RMSE: 0.028892893, Test PSNR: 18.938934326 2022-04-24 21:33:50 - Iter[137000], Epoch[000137], learning rate : 0.000227577, Train Loss: 0.136351779, Test MRAE: 0.199735105, Test RMSE: 0.027150583, Test PSNR: 18.880283356 2022-04-24 21:49:36 - Iter[138000], Epoch[000138], learning rate : 0.000225506, Train Loss: 0.136000752, Test MRAE: 0.213579893, Test RMSE: 0.030251976, Test PSNR: 19.084747314 2022-04-24 22:05:24 - Iter[139000], Epoch[000139], learning rate : 0.000223432, Train Loss: 0.135698289, Test MRAE: 0.202947915, Test RMSE: 0.027996786, Test PSNR: 19.060047150 2022-04-24 22:21:11 - Iter[140000], Epoch[000140], learning rate : 0.000221356, Train Loss: 0.135410920, Test MRAE: 0.204962358, Test RMSE: 0.028149297, Test PSNR: 19.065879822 2022-04-24 22:36:54 - Iter[141000], Epoch[000141], learning rate : 0.000219277, Train Loss: 0.135157645, Test MRAE: 0.203717798, Test RMSE: 0.029486291, Test PSNR: 19.107460022 2022-04-24 22:52:37 - Iter[142000], Epoch[000142], learning rate : 0.000217196, Train Loss: 0.134867549, Test MRAE: 0.224558994, Test RMSE: 0.031202393, Test PSNR: 19.067234039 2022-04-24 23:08:19 - Iter[143000], Epoch[000143], learning rate : 0.000215113, Train Loss: 0.134574100, Test MRAE: 0.211229146, Test RMSE: 0.031165324, Test PSNR: 19.120822906 2022-04-24 23:24:02 - Iter[144000], Epoch[000144], learning rate : 0.000213029, Train Loss: 0.134270564, Test MRAE: 0.215366766, Test RMSE: 0.030734645, Test PSNR: 19.096778870 2022-04-24 23:39:44 - Iter[145000], Epoch[000145], learning rate : 0.000210943, Train Loss: 0.134016097, Test MRAE: 0.198303372, Test RMSE: 0.027785815, Test PSNR: 19.032321930 2022-04-24 23:55:26 - Iter[146000], Epoch[000146], learning rate : 0.000208856, Train Loss: 0.133732110, Test MRAE: 0.196022749, Test RMSE: 0.029352559, Test PSNR: 19.135593414 2022-04-25 00:11:09 - Iter[147000], Epoch[000147], learning rate : 0.000206769, Train Loss: 0.133431599, Test MRAE: 0.209573299, Test RMSE: 0.029887328, Test PSNR: 19.095409393 2022-04-25 00:26:52 - Iter[148000], Epoch[000148], learning rate : 0.000204680, Train Loss: 0.133194283, Test MRAE: 0.208055094, Test RMSE: 0.028258048, Test PSNR: 18.880475998 2022-04-25 00:42:34 - Iter[149000], Epoch[000149], learning rate : 0.000202591, Train Loss: 0.132876396, Test MRAE: 0.223921135, Test RMSE: 0.030200578, Test PSNR: 18.990900040 2022-04-25 00:58:17 - Iter[150000], Epoch[000150], learning rate : 0.000200502, Train Loss: 0.132608861, Test MRAE: 0.214254171, Test RMSE: 0.028446879, Test PSNR: 18.850370407 2022-04-25 01:14:00 - Iter[151000], Epoch[000151], learning rate : 0.000198413, Train Loss: 0.132331938, Test MRAE: 0.213032275, Test RMSE: 0.028355932, Test PSNR: 18.995388031 2022-04-25 01:29:42 - Iter[152000], Epoch[000152], learning rate : 0.000196324, Train Loss: 0.132052571, Test MRAE: 0.209101513, Test RMSE: 0.029273044, Test PSNR: 19.021900177 2022-04-25 01:45:25 - Iter[153000], Epoch[000153], learning rate : 0.000194236, Train Loss: 0.131791100, Test MRAE: 0.210104674, Test RMSE: 0.029532449, Test PSNR: 19.057674408 2022-04-25 02:01:07 - Iter[154000], Epoch[000154], learning rate : 0.000192148, Train Loss: 0.131518036, Test MRAE: 0.212604702, Test RMSE: 0.028673176, Test PSNR: 18.848283768 2022-04-25 02:16:51 - Iter[155000], Epoch[000155], learning rate : 0.000190061, Train Loss: 0.131209671, Test MRAE: 0.210108310, Test RMSE: 0.030193273, Test PSNR: 18.993249893 2022-04-25 02:32:33 - Iter[156000], Epoch[000156], learning rate : 0.000187975, Train Loss: 0.130926698, Test MRAE: 0.214930877, Test RMSE: 0.029281715, Test PSNR: 18.853296280 2022-04-25 02:48:17 - Iter[157000], Epoch[000157], learning rate : 0.000185891, Train Loss: 0.130694464, Test MRAE: 0.211215034, Test RMSE: 0.029316388, Test PSNR: 18.940450668 2022-04-25 03:03:59 - Iter[158000], Epoch[000158], learning rate : 0.000183808, Train Loss: 0.130415007, Test MRAE: 0.202084705, Test RMSE: 0.028342213, Test PSNR: 19.029422760 2022-04-25 03:19:42 - Iter[159000], Epoch[000159], learning rate : 0.000181727, Train Loss: 0.130165070, Test MRAE: 0.200078711, Test RMSE: 0.028258955, Test PSNR: 19.013126373 2022-04-25 03:35:24 - Iter[160000], Epoch[000160], learning rate : 0.000179649, Train Loss: 0.129895121, Test MRAE: 0.201857880, Test RMSE: 0.027987808, Test PSNR: 19.015281677 2022-04-25 03:51:09 - Iter[161000], Epoch[000161], learning rate : 0.000177572, Train Loss: 0.129623845, Test MRAE: 0.219355986, Test RMSE: 0.030731371, Test PSNR: 19.083154678 2022-04-25 04:06:52 - Iter[162000], Epoch[000162], learning rate : 0.000175498, Train Loss: 0.129356831, Test MRAE: 0.206479296, Test RMSE: 0.028123399, Test PSNR: 19.018695831 2022-04-25 04:22:36 - Iter[163000], Epoch[000163], learning rate : 0.000173427, Train Loss: 0.129107624, Test MRAE: 0.198778838, Test RMSE: 0.028552517, Test PSNR: 19.013214111 2022-04-25 04:38:20 - Iter[164000], Epoch[000164], learning rate : 0.000171359, Train Loss: 0.128837064, Test MRAE: 0.195670083, Test RMSE: 0.027478158, Test PSNR: 18.992214203 2022-04-25 04:54:03 - Iter[165000], Epoch[000165], learning rate : 0.000169293, Train Loss: 0.128567725, Test MRAE: 0.211258903, Test RMSE: 0.029616732, Test PSNR: 19.097358704 2022-04-25 05:09:46 - Iter[166000], Epoch[000166], learning rate : 0.000167232, Train Loss: 0.128432900, Test MRAE: 0.191220835, Test RMSE: 0.027260069, Test PSNR: 18.941967010 2022-04-25 05:25:28 - Iter[167000], Epoch[000167], learning rate : 0.000165174, Train Loss: 0.128152385, Test MRAE: 0.202767581, Test RMSE: 0.028542725, Test PSNR: 18.998825073 2022-04-25 05:41:12 - Iter[168000], Epoch[000168], learning rate : 0.000163119, Train Loss: 0.127891243, Test MRAE: 0.204690695, Test RMSE: 0.028364403, Test PSNR: 18.970281601 2022-04-25 05:56:54 - Iter[169000], Epoch[000169], learning rate : 0.000161069, Train Loss: 0.127617165, Test MRAE: 0.203078985, Test RMSE: 0.028361408, Test PSNR: 19.034769058 2022-04-25 06:12:38 - Iter[170000], Epoch[000170], learning rate : 0.000159024, Train Loss: 0.127377793, Test MRAE: 0.206805512, Test RMSE: 0.030018816, Test PSNR: 19.061866760 2022-04-25 06:28:23 - Iter[171000], Epoch[000171], learning rate : 0.000156982, Train Loss: 0.127140224, Test MRAE: 0.208625391, Test RMSE: 0.028507199, Test PSNR: 18.989234924 2022-04-25 06:44:06 - Iter[172000], Epoch[000172], learning rate : 0.000154946, Train Loss: 0.126904130, Test MRAE: 0.210705116, Test RMSE: 0.029785754, Test PSNR: 18.994005203 2022-04-25 06:59:49 - Iter[173000], Epoch[000173], learning rate : 0.000152915, Train Loss: 0.126656905, Test MRAE: 0.210085228, Test RMSE: 0.029623386, Test PSNR: 19.080978394 2022-04-25 07:15:31 - Iter[174000], Epoch[000174], learning rate : 0.000150888, Train Loss: 0.126484647, Test MRAE: 0.215656236, Test RMSE: 0.030040557, Test PSNR: 19.056533813 2022-04-25 07:31:13 - Iter[175000], Epoch[000175], learning rate : 0.000148868, Train Loss: 0.126255050, Test MRAE: 0.216682300, Test RMSE: 0.029841734, Test PSNR: 18.989347458 2022-04-25 07:46:56 - Iter[176000], Epoch[000176], learning rate : 0.000146853, Train Loss: 0.126007512, Test MRAE: 0.229140550, Test RMSE: 0.031297620, Test PSNR: 18.960144043 2022-04-25 08:02:38 - Iter[177000], Epoch[000177], learning rate : 0.000144843, Train Loss: 0.125766575, Test MRAE: 0.218668729, Test RMSE: 0.031003775, Test PSNR: 19.038333893 2022-04-25 08:18:22 - Iter[178000], Epoch[000178], learning rate : 0.000142840, Train Loss: 0.125532001, Test MRAE: 0.199688122, Test RMSE: 0.028291211, Test PSNR: 18.974950790 2022-04-25 08:34:09 - Iter[179000], Epoch[000179], learning rate : 0.000140843, Train Loss: 0.125298381, Test MRAE: 0.207475007, Test RMSE: 0.029013749, Test PSNR: 18.985069275 2022-04-25 08:49:51 - Iter[180000], Epoch[000180], learning rate : 0.000138853, Train Loss: 0.125062704, Test MRAE: 0.206389904, Test RMSE: 0.029368754, Test PSNR: 19.005840302 2022-04-25 09:05:36 - Iter[181000], Epoch[000181], learning rate : 0.000136870, Train Loss: 0.124840498, Test MRAE: 0.210619673, Test RMSE: 0.029867401, Test PSNR: 18.961513519 2022-04-25 09:21:18 - Iter[182000], Epoch[000182], learning rate : 0.000134893, Train Loss: 0.124605544, Test MRAE: 0.220269203, Test RMSE: 0.032067895, Test PSNR: 19.168319702 2022-04-25 09:37:01 - Iter[183000], Epoch[000183], learning rate : 0.000132924, Train Loss: 0.124368042, Test MRAE: 0.191453904, Test RMSE: 0.027029432, Test PSNR: 19.038944244 2022-04-25 09:52:47 - Iter[184000], Epoch[000184], learning rate : 0.000130962, Train Loss: 0.124137081, Test MRAE: 0.206539020, Test RMSE: 0.029571155, Test PSNR: 19.048000336 2022-04-25 10:08:30 - Iter[185000], Epoch[000185], learning rate : 0.000129008, Train Loss: 0.123918340, Test MRAE: 0.194905445, Test RMSE: 0.027700884, Test PSNR: 19.005758286 2022-04-25 10:24:14 - Iter[186000], Epoch[000186], learning rate : 0.000127061, Train Loss: 0.123697750, Test MRAE: 0.205566064, Test RMSE: 0.028691338, Test PSNR: 18.983287811 2022-04-25 10:39:56 - Iter[187000], Epoch[000187], learning rate : 0.000125123, Train Loss: 0.123466983, Test MRAE: 0.201362506, Test RMSE: 0.028842332, Test PSNR: 19.055122375 2022-04-25 10:55:39 - Iter[188000], Epoch[000188], learning rate : 0.000123193, Train Loss: 0.123243488, Test MRAE: 0.206426546, Test RMSE: 0.029135758, Test PSNR: 19.006008148 2022-04-25 11:11:23 - Iter[189000], Epoch[000189], learning rate : 0.000121271, Train Loss: 0.123076245, Test MRAE: 0.207691565, Test RMSE: 0.029229913, Test PSNR: 19.043151855 2022-04-25 11:27:06 - Iter[190000], Epoch[000190], learning rate : 0.000119358, Train Loss: 0.122856230, Test MRAE: 0.206229493, Test RMSE: 0.028870497, Test PSNR: 18.966106415 2022-04-25 11:42:49 - Iter[191000], Epoch[000191], learning rate : 0.000117454, Train Loss: 0.122638315, Test MRAE: 0.205771387, Test RMSE: 0.028803570, Test PSNR: 18.946226120 2022-04-25 11:58:32 - Iter[192000], Epoch[000192], learning rate : 0.000115559, Train Loss: 0.122420080, Test MRAE: 0.212724239, Test RMSE: 0.030358646, Test PSNR: 19.106163025 2022-04-25 12:14:15 - Iter[193000], Epoch[000193], learning rate : 0.000113673, Train Loss: 0.122199051, Test MRAE: 0.214906707, Test RMSE: 0.030220853, Test PSNR: 19.064342499 2022-04-25 12:30:00 - Iter[194000], Epoch[000194], learning rate : 0.000111797, Train Loss: 0.121984355, Test MRAE: 0.198100254, Test RMSE: 0.028148143, Test PSNR: 18.971063614 2022-04-25 12:45:43 - Iter[195000], Epoch[000195], learning rate : 0.000109931, Train Loss: 0.121781416, Test MRAE: 0.198713332, Test RMSE: 0.028723657, Test PSNR: 19.015140533 2022-04-25 13:01:27 - Iter[196000], Epoch[000196], learning rate : 0.000108074, Train Loss: 0.121572427, Test MRAE: 0.197447866, Test RMSE: 0.028784798, Test PSNR: 18.987123489 2022-04-25 13:17:11 - Iter[197000], Epoch[000197], learning rate : 0.000106228, Train Loss: 0.121370003, Test MRAE: 0.205944434, Test RMSE: 0.029543681, Test PSNR: 19.049442291 2022-04-25 13:32:58 - Iter[198000], Epoch[000198], learning rate : 0.000104392, Train Loss: 0.101264954, Test MRAE: 0.202921927, Test RMSE: 0.029191626, Test PSNR: 19.073200226 2022-04-25 13:48:45 - Iter[199000], Epoch[000199], learning rate : 0.000102567, Train Loss: 0.099872775, Test MRAE: 0.213733122, Test RMSE: 0.030334827, Test PSNR: 19.099483490 2022-04-25 14:04:30 - Iter[200000], Epoch[000200], learning rate : 0.000100752, Train Loss: 0.099436894, Test MRAE: 0.207790688, Test RMSE: 0.029580811, Test PSNR: 19.065809250 2022-04-25 14:20:13 - Iter[201000], Epoch[000201], learning rate : 0.000098948, Train Loss: 0.099523008, Test MRAE: 0.206653327, Test RMSE: 0.029507691, Test PSNR: 19.007652283 2022-04-25 14:35:57 - Iter[202000], Epoch[000202], learning rate : 0.000097155, Train Loss: 0.099680349, Test MRAE: 0.209775746, Test RMSE: 0.029743711, Test PSNR: 19.063774109 2022-04-25 14:51:41 - Iter[203000], Epoch[000203], learning rate : 0.000095374, Train Loss: 0.099537373, Test MRAE: 0.205857038, Test RMSE: 0.029820710, Test PSNR: 19.054101944 2022-04-25 15:07:24 - Iter[204000], Epoch[000204], learning rate : 0.000093604, Train Loss: 0.099403314, Test MRAE: 0.201450929, Test RMSE: 0.028907372, Test PSNR: 18.933977127 2022-04-25 15:23:07 - Iter[205000], Epoch[000205], learning rate : 0.000091846, Train Loss: 0.099265657, Test MRAE: 0.202834055, Test RMSE: 0.029253015, Test PSNR: 18.995012283 2022-04-25 15:38:50 - Iter[206000], Epoch[000206], learning rate : 0.000090100, Train Loss: 0.099154606, Test MRAE: 0.200972140, Test RMSE: 0.028723257, Test PSNR: 18.992853165 2022-04-25 15:54:34 - Iter[207000], Epoch[000207], learning rate : 0.000088366, Train Loss: 0.099067487, Test MRAE: 0.206179917, Test RMSE: 0.029294010, Test PSNR: 18.984216690 2022-04-25 16:10:17 - Iter[208000], Epoch[000208], learning rate : 0.000086644, Train Loss: 0.098957688, Test MRAE: 0.201780394, Test RMSE: 0.028410414, Test PSNR: 19.005832672 2022-04-25 16:26:03 - Iter[209000], Epoch[000209], learning rate : 0.000084935, Train Loss: 0.098789133, Test MRAE: 0.209743783, Test RMSE: 0.029414069, Test PSNR: 18.970657349 2022-04-25 16:41:47 - Iter[210000], Epoch[000210], learning rate : 0.000083239, Train Loss: 0.098655112, Test MRAE: 0.200564787, Test RMSE: 0.028275695, Test PSNR: 19.033458710 2022-04-25 16:57:32 - Iter[211000], Epoch[000211], learning rate : 0.000081555, Train Loss: 0.098488487, Test MRAE: 0.199889153, Test RMSE: 0.028281061, Test PSNR: 18.950765610 2022-04-25 17:13:15 - Iter[212000], Epoch[000212], learning rate : 0.000079884, Train Loss: 0.098348401, Test MRAE: 0.200752750, Test RMSE: 0.027866315, Test PSNR: 18.983562469 2022-04-25 17:28:57 - Iter[213000], Epoch[000213], learning rate : 0.000078227, Train Loss: 0.098229684, Test MRAE: 0.201839328, Test RMSE: 0.028493920, Test PSNR: 18.957841873 2022-04-25 17:44:40 - Iter[214000], Epoch[000214], learning rate : 0.000076583, Train Loss: 0.098102383, Test MRAE: 0.204463542, Test RMSE: 0.028648939, Test PSNR: 18.996351242 2022-04-25 18:00:22 - Iter[215000], Epoch[000215], learning rate : 0.000074952, Train Loss: 0.097989276, Test MRAE: 0.201856405, Test RMSE: 0.028056156, Test PSNR: 18.929220200 2022-04-25 18:16:08 - Iter[216000], Epoch[000216], learning rate : 0.000073336, Train Loss: 0.097863868, Test MRAE: 0.210833490, Test RMSE: 0.030079007, Test PSNR: 19.039060593 2022-04-25 18:31:50 - Iter[217000], Epoch[000217], learning rate : 0.000071733, Train Loss: 0.097764373, Test MRAE: 0.199148744, Test RMSE: 0.027379682, Test PSNR: 18.915548325 2022-04-25 18:47:33 - Iter[218000], Epoch[000218], learning rate : 0.000070144, Train Loss: 0.097662948, Test MRAE: 0.211309999, Test RMSE: 0.029880499, Test PSNR: 19.041957855 2022-04-25 19:03:16 - Iter[219000], Epoch[000219], learning rate : 0.000068570, Train Loss: 0.097545773, Test MRAE: 0.202322707, Test RMSE: 0.029350601, Test PSNR: 19.032449722 2022-04-25 19:19:05 - Iter[220000], Epoch[000220], learning rate : 0.000067010, Train Loss: 0.097439609, Test MRAE: 0.201889187, Test RMSE: 0.028135814, Test PSNR: 19.002304077 2022-04-25 19:34:58 - Iter[221000], Epoch[000221], learning rate : 0.000065465, Train Loss: 0.097333550, Test MRAE: 0.205386743, Test RMSE: 0.029404126, Test PSNR: 19.006492615 2022-04-25 19:50:42 - Iter[222000], Epoch[000222], learning rate : 0.000063934, Train Loss: 0.097243309, Test MRAE: 0.200861678, Test RMSE: 0.028495271, Test PSNR: 19.029602051 2022-04-25 20:06:24 - Iter[223000], Epoch[000223], learning rate : 0.000062419, Train Loss: 0.097154871, Test MRAE: 0.209499523, Test RMSE: 0.029613551, Test PSNR: 19.025751114 2022-04-25 20:22:07 - Iter[224000], Epoch[000224], learning rate : 0.000060919, Train Loss: 0.097066604, Test MRAE: 0.216345757, Test RMSE: 0.030452432, Test PSNR: 19.079719543 2022-04-25 20:37:50 - Iter[225000], Epoch[000225], learning rate : 0.000059434, Train Loss: 0.096955277, Test MRAE: 0.207744464, Test RMSE: 0.029243071, Test PSNR: 19.021963120 2022-04-25 20:53:34 - Iter[226000], Epoch[000226], learning rate : 0.000057964, Train Loss: 0.096838258, Test MRAE: 0.199603081, Test RMSE: 0.028276462, Test PSNR: 19.053121567 2022-04-25 21:09:17 - Iter[227000], Epoch[000227], learning rate : 0.000056510, Train Loss: 0.096710235, Test MRAE: 0.197099164, Test RMSE: 0.027980030, Test PSNR: 18.989082336 2022-04-25 21:24:59 - Iter[228000], Epoch[000228], learning rate : 0.000055072, Train Loss: 0.096593060, Test MRAE: 0.192215458, Test RMSE: 0.027695557, Test PSNR: 19.051717758 2022-04-25 21:40:41 - Iter[229000], Epoch[000229], learning rate : 0.000053650, Train Loss: 0.096480407, Test MRAE: 0.194151208, Test RMSE: 0.028371762, Test PSNR: 19.065599442 2022-04-25 21:56:29 - Iter[230000], Epoch[000230], learning rate : 0.000052244, Train Loss: 0.096383542, Test MRAE: 0.203011096, Test RMSE: 0.029121868, Test PSNR: 19.086605072 2022-04-25 22:12:12 - Iter[231000], Epoch[000231], learning rate : 0.000050854, Train Loss: 0.096299224, Test MRAE: 0.201670945, Test RMSE: 0.028482547, Test PSNR: 19.051557541 2022-04-25 22:27:55 - Iter[232000], Epoch[000232], learning rate : 0.000049481, Train Loss: 0.096198440, Test MRAE: 0.196830481, Test RMSE: 0.028049719, Test PSNR: 19.039011002 2022-04-25 22:43:37 - Iter[233000], Epoch[000233], learning rate : 0.000048124, Train Loss: 0.096105792, Test MRAE: 0.194558725, Test RMSE: 0.027864750, Test PSNR: 19.071357727 2022-04-25 22:59:20 - Iter[234000], Epoch[000234], learning rate : 0.000046784, Train Loss: 0.096012853, Test MRAE: 0.200894669, Test RMSE: 0.028464397, Test PSNR: 19.031705856 2022-04-25 23:15:02 - Iter[235000], Epoch[000235], learning rate : 0.000045461, Train Loss: 0.095911391, Test MRAE: 0.203084469, Test RMSE: 0.029015776, Test PSNR: 19.042705536 2022-04-25 23:30:45 - Iter[236000], Epoch[000236], learning rate : 0.000044154, Train Loss: 0.095807374, Test MRAE: 0.205492422, Test RMSE: 0.029720480, Test PSNR: 19.064033508 2022-04-25 23:46:27 - Iter[237000], Epoch[000237], learning rate : 0.000042865, Train Loss: 0.095718473, Test MRAE: 0.200766295, Test RMSE: 0.029130232, Test PSNR: 19.070289612 2022-04-26 00:02:16 - Iter[238000], Epoch[000238], learning rate : 0.000041594, Train Loss: 0.095622800, Test MRAE: 0.201171368, Test RMSE: 0.028642515, Test PSNR: 19.076307297 2022-04-26 00:17:58 - Iter[239000], Epoch[000239], learning rate : 0.000040339, Train Loss: 0.095527329, Test MRAE: 0.207034603, Test RMSE: 0.030098038, Test PSNR: 19.122308731 2022-04-26 00:33:41 - Iter[240000], Epoch[000240], learning rate : 0.000039102, Train Loss: 0.095431149, Test MRAE: 0.200998485, Test RMSE: 0.028509894, Test PSNR: 19.048767090 2022-04-26 00:49:25 - Iter[241000], Epoch[000241], learning rate : 0.000037883, Train Loss: 0.095341481, Test MRAE: 0.197351202, Test RMSE: 0.027994553, Test PSNR: 19.053705215 2022-04-26 01:05:07 - Iter[242000], Epoch[000242], learning rate : 0.000036682, Train Loss: 0.095249861, Test MRAE: 0.205203146, Test RMSE: 0.029421324, Test PSNR: 19.068685532 2022-04-26 01:20:51 - Iter[243000], Epoch[000243], learning rate : 0.000035499, Train Loss: 0.095152855, Test MRAE: 0.202429041, Test RMSE: 0.028778778, Test PSNR: 19.062547684 2022-04-26 01:36:39 - Iter[244000], Epoch[000244], learning rate : 0.000034333, Train Loss: 0.095076188, Test MRAE: 0.199525461, Test RMSE: 0.028577324, Test PSNR: 19.065444946 2022-04-26 01:52:28 - Iter[245000], Epoch[000245], learning rate : 0.000033186, Train Loss: 0.094988756, Test MRAE: 0.193906844, Test RMSE: 0.027360469, Test PSNR: 19.035274506 2022-04-26 02:08:13 - Iter[246000], Epoch[000246], learning rate : 0.000032058, Train Loss: 0.094907209, Test MRAE: 0.194509611, Test RMSE: 0.027639085, Test PSNR: 19.038093567 2022-04-26 02:23:58 - Iter[247000], Epoch[000247], learning rate : 0.000030948, Train Loss: 0.094821684, Test MRAE: 0.205400094, Test RMSE: 0.029638980, Test PSNR: 19.069122314 2022-04-26 02:39:41 - Iter[248000], Epoch[000248], learning rate : 0.000029856, Train Loss: 0.094757117, Test MRAE: 0.200400651, Test RMSE: 0.028777106, Test PSNR: 19.065080643 2022-04-26 02:55:26 - Iter[249000], Epoch[000249], learning rate : 0.000028783, Train Loss: 0.094666235, Test MRAE: 0.196040452, Test RMSE: 0.027733754, Test PSNR: 19.004329681 2022-04-26 03:11:09 - Iter[250000], Epoch[000250], learning rate : 0.000027729, Train Loss: 0.094580248, Test MRAE: 0.198402479, Test RMSE: 0.028335137, Test PSNR: 19.012487411 2022-04-26 03:26:51 - Iter[251000], Epoch[000251], learning rate : 0.000026694, Train Loss: 0.094503880, Test MRAE: 0.198768482, Test RMSE: 0.027874328, Test PSNR: 18.994089127 2022-04-26 03:42:33 - Iter[252000], Epoch[000252], learning rate : 0.000025678, Train Loss: 0.094424129, Test MRAE: 0.204637840, Test RMSE: 0.028599529, Test PSNR: 19.048522949 2022-04-26 03:58:17 - Iter[253000], Epoch[000253], learning rate : 0.000024681, Train Loss: 0.094356239, Test MRAE: 0.197843701, Test RMSE: 0.027720425, Test PSNR: 19.019216537 2022-04-26 04:13:59 - Iter[254000], Epoch[000254], learning rate : 0.000023703, Train Loss: 0.094279803, Test MRAE: 0.195381641, Test RMSE: 0.027696660, Test PSNR: 19.030433655 2022-04-26 04:29:44 - Iter[255000], Epoch[000255], learning rate : 0.000022745, Train Loss: 0.094200850, Test MRAE: 0.201875895, Test RMSE: 0.028370189, Test PSNR: 19.016584396 2022-04-26 04:45:26 - Iter[256000], Epoch[000256], learning rate : 0.000021806, Train Loss: 0.094126128, Test MRAE: 0.203219756, Test RMSE: 0.028974859, Test PSNR: 19.087203979 2022-04-26 05:01:09 - Iter[257000], Epoch[000257], learning rate : 0.000020887, Train Loss: 0.094045103, Test MRAE: 0.194824770, Test RMSE: 0.027946815, Test PSNR: 19.061742783 2022-04-26 05:16:51 - Iter[258000], Epoch[000258], learning rate : 0.000019988, Train Loss: 0.093975663, Test MRAE: 0.198491260, Test RMSE: 0.027982576, Test PSNR: 19.010288239 2022-04-26 05:32:35 - Iter[259000], Epoch[000259], learning rate : 0.000019108, Train Loss: 0.093899436, Test MRAE: 0.201042473, Test RMSE: 0.028101049, Test PSNR: 19.056558609 2022-04-26 05:48:19 - Iter[260000], Epoch[000260], learning rate : 0.000018249, Train Loss: 0.093830153, Test MRAE: 0.201697826, Test RMSE: 0.028358519, Test PSNR: 19.034065247 2022-04-26 06:04:01 - Iter[261000], Epoch[000261], learning rate : 0.000017409, Train Loss: 0.093763351, Test MRAE: 0.200538099, Test RMSE: 0.028340435, Test PSNR: 19.020803452 2022-04-26 06:19:44 - Iter[262000], Epoch[000262], learning rate : 0.000016589, Train Loss: 0.093695477, Test MRAE: 0.202214047, Test RMSE: 0.028496150, Test PSNR: 19.040538788 2022-04-26 06:35:27 - Iter[263000], Epoch[000263], learning rate : 0.000015790, Train Loss: 0.093628220, Test MRAE: 0.199471787, Test RMSE: 0.028235294, Test PSNR: 19.043592453 2022-04-26 06:51:13 - Iter[264000], Epoch[000264], learning rate : 0.000015010, Train Loss: 0.093556948, Test MRAE: 0.200963169, Test RMSE: 0.028326141, Test PSNR: 19.030656815 2022-04-26 07:07:01 - Iter[265000], Epoch[000265], learning rate : 0.000014251, Train Loss: 0.093489110, Test MRAE: 0.197174221, Test RMSE: 0.027766764, Test PSNR: 19.029552460 2022-04-26 07:22:46 - Iter[266000], Epoch[000266], learning rate : 0.000013513, Train Loss: 0.093422472, Test MRAE: 0.198560372, Test RMSE: 0.027845098, Test PSNR: 19.010746002 2022-04-26 07:38:30 - Iter[267000], Epoch[000267], learning rate : 0.000012795, Train Loss: 0.093358673, Test MRAE: 0.198623791, Test RMSE: 0.028192090, Test PSNR: 19.046485901 2022-04-26 07:54:13 - Iter[268000], Epoch[000268], learning rate : 0.000012098, Train Loss: 0.093292192, Test MRAE: 0.197530746, Test RMSE: 0.028094612, Test PSNR: 19.029310226 2022-04-26 08:09:59 - Iter[269000], Epoch[000269], learning rate : 0.000011421, Train Loss: 0.093230203, Test MRAE: 0.199864402, Test RMSE: 0.028338477, Test PSNR: 19.030046463 2022-04-26 08:25:42 - Iter[270000], Epoch[000270], learning rate : 0.000010765, Train Loss: 0.093164317, Test MRAE: 0.196141958, Test RMSE: 0.027876040, Test PSNR: 19.026119232 2022-04-26 08:41:25 - Iter[271000], Epoch[000271], learning rate : 0.000010130, Train Loss: 0.093100064, Test MRAE: 0.199385583, Test RMSE: 0.027962262, Test PSNR: 19.018707275 2022-04-26 08:57:08 - Iter[272000], Epoch[000272], learning rate : 0.000009515, Train Loss: 0.093038529, Test MRAE: 0.203236878, Test RMSE: 0.028776500, Test PSNR: 19.067478180 2022-04-26 09:12:51 - Iter[273000], Epoch[000273], learning rate : 0.000008922, Train Loss: 0.092979573, Test MRAE: 0.201909021, Test RMSE: 0.028371701, Test PSNR: 19.025129318 2022-04-26 09:28:33 - Iter[274000], Epoch[000274], learning rate : 0.000008350, Train Loss: 0.092915766, Test MRAE: 0.199700192, Test RMSE: 0.028185047, Test PSNR: 19.029628754 2022-04-26 09:44:18 - Iter[275000], Epoch[000275], learning rate : 0.000007798, Train Loss: 0.092863925, Test MRAE: 0.199149221, Test RMSE: 0.028089032, Test PSNR: 19.003961563 2022-04-26 10:00:00 - Iter[276000], Epoch[000276], learning rate : 0.000007268, Train Loss: 0.092805415, Test MRAE: 0.196530923, Test RMSE: 0.027886262, Test PSNR: 19.012132645 2022-04-26 10:15:43 - Iter[277000], Epoch[000277], learning rate : 0.000006759, Train Loss: 0.092753656, Test MRAE: 0.197557405, Test RMSE: 0.028128177, Test PSNR: 19.038337708 2022-04-26 10:31:26 - Iter[278000], Epoch[000278], learning rate : 0.000006271, Train Loss: 0.092694871, Test MRAE: 0.197338894, Test RMSE: 0.027873993, Test PSNR: 19.016971588 2022-04-26 10:47:11 - Iter[279000], Epoch[000279], learning rate : 0.000005805, Train Loss: 0.092640802, Test MRAE: 0.199877754, Test RMSE: 0.028381795, Test PSNR: 19.056432724 2022-04-26 11:02:55 - Iter[280000], Epoch[000280], learning rate : 0.000005360, Train Loss: 0.092587359, Test MRAE: 0.198997468, Test RMSE: 0.028246677, Test PSNR: 19.035556793 2022-04-26 11:18:37 - Iter[281000], Epoch[000281], learning rate : 0.000004936, Train Loss: 0.092534050, Test MRAE: 0.198044509, Test RMSE: 0.027995434, Test PSNR: 19.035224915 2022-04-26 11:34:25 - Iter[282000], Epoch[000282], learning rate : 0.000004534, Train Loss: 0.092481837, Test MRAE: 0.199215770, Test RMSE: 0.028157072, Test PSNR: 19.028303146 2022-04-26 11:50:09 - Iter[283000], Epoch[000283], learning rate : 0.000004153, Train Loss: 0.092428215, Test MRAE: 0.199768141, Test RMSE: 0.028234014, Test PSNR: 19.032674789 2022-04-26 12:05:52 - Iter[284000], Epoch[000284], learning rate : 0.000003794, Train Loss: 0.092372514, Test MRAE: 0.200132683, Test RMSE: 0.028386112, Test PSNR: 19.050033569 2022-04-26 12:21:34 - Iter[285000], Epoch[000285], learning rate : 0.000003457, Train Loss: 0.092325777, Test MRAE: 0.200290814, Test RMSE: 0.028357379, Test PSNR: 19.042633057 2022-04-26 12:37:16 - Iter[286000], Epoch[000286], learning rate : 0.000003140, Train Loss: 0.092276767, Test MRAE: 0.199622378, Test RMSE: 0.028264590, Test PSNR: 19.027994156 2022-04-26 12:52:59 - Iter[287000], Epoch[000287], learning rate : 0.000002846, Train Loss: 0.092229761, Test MRAE: 0.200462580, Test RMSE: 0.028426396, Test PSNR: 19.040649414 2022-04-26 13:08:45 - Iter[288000], Epoch[000288], learning rate : 0.000002573, Train Loss: 0.092182025, Test MRAE: 0.199467480, Test RMSE: 0.028248770, Test PSNR: 19.024812698 2022-04-26 13:24:28 - Iter[289000], Epoch[000289], learning rate : 0.000002322, Train Loss: 0.092134498, Test MRAE: 0.199127629, Test RMSE: 0.028083034, Test PSNR: 19.012405396 2022-04-26 13:40:10 - Iter[290000], Epoch[000290], learning rate : 0.000002093, Train Loss: 0.092089295, Test MRAE: 0.200336218, Test RMSE: 0.028368756, Test PSNR: 19.028881073 2022-04-26 13:55:54 - Iter[291000], Epoch[000291], learning rate : 0.000001886, Train Loss: 0.092047319, Test MRAE: 0.198826462, Test RMSE: 0.028130181, Test PSNR: 19.023952484 2022-04-26 14:11:39 - Iter[292000], Epoch[000292], learning rate : 0.000001700, Train Loss: 0.092004254, Test MRAE: 0.200323239, Test RMSE: 0.028369704, Test PSNR: 19.032167435 2022-04-26 14:27:22 - Iter[293000], Epoch[000293], learning rate : 0.000001536, Train Loss: 0.091962963, Test MRAE: 0.198782578, Test RMSE: 0.028092362, Test PSNR: 19.022401810 2022-04-26 14:43:07 - Iter[294000], Epoch[000294], learning rate : 0.000001394, Train Loss: 0.091920927, Test MRAE: 0.199555337, Test RMSE: 0.028241893, Test PSNR: 19.032341003 2022-04-26 14:58:52 - Iter[295000], Epoch[000295], learning rate : 0.000001274, Train Loss: 0.091873795, Test MRAE: 0.200663373, Test RMSE: 0.028400686, Test PSNR: 19.039638519 2022-04-26 15:14:36 - Iter[296000], Epoch[000296], learning rate : 0.000001175, Train Loss: 0.091835335, Test MRAE: 0.199872047, Test RMSE: 0.028326735, Test PSNR: 19.036495209 2022-04-26 15:30:21 - Iter[297000], Epoch[000297], learning rate : 0.000001099, Train Loss: 0.087091446, Test MRAE: 0.200016499, Test RMSE: 0.028326962, Test PSNR: 19.033159256 2022-04-26 15:46:09 - Iter[298000], Epoch[000298], learning rate : 0.000001044, Train Loss: 0.087477118, Test MRAE: 0.199742630, Test RMSE: 0.028276479, Test PSNR: 19.029752731 2022-04-26 16:01:52 - Iter[299000], Epoch[000299], learning rate : 0.000001011, Train Loss: 0.087522693, Test MRAE: 0.200146362, Test RMSE: 0.028322442, Test PSNR: 19.031415939 2022-04-26 16:17:34 - Iter[300000], Epoch[000300], learning rate : 0.000001000, Train Loss: 0.087497599, Test MRAE: 0.200221285, Test RMSE: 0.028348781, Test PSNR: 19.032201767 2022-04-26 16:33:17 - Iter[301000], Epoch[000301], learning rate : 0.000001011, Train Loss: 0.087575421, Test MRAE: 0.200616375, Test RMSE: 0.028419724, Test PSNR: 19.031145096 2022-04-26 16:49:01 - Iter[302000], Epoch[000302], learning rate : 0.000001044, Train Loss: 0.087664612, Test MRAE: 0.199603871, Test RMSE: 0.028222578, Test PSNR: 19.027841568 2022-04-26 17:04:44 - Iter[303000], Epoch[000303], learning rate : 0.000001098, Train Loss: 0.087709896, Test MRAE: 0.199289769, Test RMSE: 0.028206011, Test PSNR: 19.033077240 2022-04-26 17:20:27 - Iter[304000], Epoch[000304], learning rate : 0.000001175, Train Loss: 0.087701336, Test MRAE: 0.198439360, Test RMSE: 0.028052971, Test PSNR: 19.023553848 2022-04-26 17:36:09 - Iter[305000], Epoch[000305], learning rate : 0.000001273, Train Loss: 0.087724887, Test MRAE: 0.199188337, Test RMSE: 0.028147060, Test PSNR: 19.024749756 2022-04-26 17:51:52 - Iter[306000], Epoch[000306], learning rate : 0.000001394, Train Loss: 0.087697797, Test MRAE: 0.198744774, Test RMSE: 0.028123770, Test PSNR: 19.021911621 2022-04-26 18:07:40 - Iter[307000], Epoch[000307], learning rate : 0.000001536, Train Loss: 0.087724082, Test MRAE: 0.198784173, Test RMSE: 0.028199598, Test PSNR: 19.027591705 2022-04-26 18:23:26 - Iter[308000], Epoch[000308], learning rate : 0.000001699, Train Loss: 0.087754786, Test MRAE: 0.200074822, Test RMSE: 0.028291941, Test PSNR: 19.026384354 2022-04-26 18:39:09 - Iter[309000], Epoch[000309], learning rate : 0.000001885, Train Loss: 0.087743573, Test MRAE: 0.199570522, Test RMSE: 0.028250307, Test PSNR: 19.019496918 2022-04-26 18:54:53 - Iter[310000], Epoch[000310], learning rate : 0.000002093, Train Loss: 0.087756194, Test MRAE: 0.200466946, Test RMSE: 0.028466217, Test PSNR: 19.043897629 2022-04-26 19:10:36 - Iter[311000], Epoch[000311], learning rate : 0.000002322, Train Loss: 0.087725841, Test MRAE: 0.200274870, Test RMSE: 0.028353559, Test PSNR: 19.034431458 2022-04-26 19:26:18 - Iter[312000], Epoch[000312], learning rate : 0.000002573, Train Loss: 0.087727122, Test MRAE: 0.199417949, Test RMSE: 0.028240953, Test PSNR: 19.032062531 2022-04-26 19:42:00 - Iter[313000], Epoch[000313], learning rate : 0.000002846, Train Loss: 0.087730475, Test MRAE: 0.200596809, Test RMSE: 0.028354665, Test PSNR: 19.033941269 2022-04-26 19:57:43 - Iter[314000], Epoch[000314], learning rate : 0.000003140, Train Loss: 0.087749563, Test MRAE: 0.201247305, Test RMSE: 0.028494956, Test PSNR: 19.038160324 2022-04-26 20:13:26 - Iter[315000], Epoch[000315], learning rate : 0.000003456, Train Loss: 0.087754004, Test MRAE: 0.200625256, Test RMSE: 0.028311061, Test PSNR: 19.022411346 2022-04-26 20:29:09 - Iter[316000], Epoch[000316], learning rate : 0.000003793, Train Loss: 0.087767355, Test MRAE: 0.199434310, Test RMSE: 0.028191015, Test PSNR: 19.015176773 2022-04-26 20:44:59 - Iter[317000], Epoch[000317], learning rate : 0.000004153, Train Loss: 0.087786980, Test MRAE: 0.201855198, Test RMSE: 0.028614521, Test PSNR: 19.041559219 2022-04-26 21:00:41 - Iter[318000], Epoch[000318], learning rate : 0.000004533, Train Loss: 0.087780491, Test MRAE: 0.200222835, Test RMSE: 0.028176807, Test PSNR: 19.015808105 2022-04-26 21:16:24 - Iter[319000], Epoch[000319], learning rate : 0.000004935, Train Loss: 0.087792754, Test MRAE: 0.199134097, Test RMSE: 0.028215753, Test PSNR: 19.028671265 2022-04-26 21:32:06 - Iter[320000], Epoch[000320], learning rate : 0.000005359, Train Loss: 0.087798759, Test MRAE: 0.200270444, Test RMSE: 0.028273745, Test PSNR: 19.037963867 2022-04-26 21:47:48 - Iter[321000], Epoch[000321], learning rate : 0.000005804, Train Loss: 0.087813973, Test MRAE: 0.200781301, Test RMSE: 0.028401980, Test PSNR: 19.028430939 2022-04-26 22:03:33 - Iter[322000], Epoch[000322], learning rate : 0.000006271, Train Loss: 0.087820567, Test MRAE: 0.199449122, Test RMSE: 0.028095623, Test PSNR: 19.015094757 2022-04-26 22:19:15 - Iter[323000], Epoch[000323], learning rate : 0.000006758, Train Loss: 0.087827265, Test MRAE: 0.201382726, Test RMSE: 0.028672023, Test PSNR: 19.032772064 2022-04-26 22:34:57 - Iter[324000], Epoch[000324], learning rate : 0.000007267, Train Loss: 0.087820075, Test MRAE: 0.199561149, Test RMSE: 0.028316684, Test PSNR: 19.027658463 2022-04-26 22:50:42 - Iter[325000], Epoch[000325], learning rate : 0.000007797, Train Loss: 0.087809280, Test MRAE: 0.202198595, Test RMSE: 0.028633879, Test PSNR: 19.034700394 2022-04-26 23:06:24 - Iter[326000], Epoch[000326], learning rate : 0.000008349, Train Loss: 0.087817691, Test MRAE:

    opened by xcyquan 10
  • MST++和Restormer的区别

    MST++和Restormer的区别

    MST++仅使用1/10的参数量取得比Restormer更好的效果,我看了下代码觉得二者的模型架构差不多,请问一下是哪一步修改产生如此巨大的改进吗?

    我对比了下代码,区别主要有

    1. Restormer使用门控FFN,MST++取消了门控
    2. 和Restormer相比,MST++去掉了通道注意力模块(Spectral-wise Multi-head Self-Attention) 中产生Q,K,V向量的DW Conv
    3. 和Restormer相比,MST++在通道注意力模块增加了卷积位置编码
    4. MST++去掉了通道注意力模块的LayerNorm
    opened by madfff 7
  • Loss very quickly reduces but PSNR and RMSE doesn't improve

    Loss very quickly reduces but PSNR and RMSE doesn't improve

    Hi authors!

    I appreciate your work done. It's quite inspiring!

    I am trying to train MST++ from scratch. I just followed the commands mentioned but even after training for long time my RMSE, MRAE and PSNR doesn't improve. I have even tried using recommended environment settings. I am using same settings (i.e. batch size, lr scheduling etc.).

    2022-08-23 10:03:27 - Iter[001000], Epoch[000001], learning rate : 0.000399989, Train Loss: 0.345733970, Test MRAE: 5.343356609, Test RMSE: 0.567134500, Test PSNR: 16.002279282 2022-08-23 10:20:02 - Iter[002000], Epoch[000002], learning rate : 0.000399956, Train Loss: 0.242199719, Test MRAE: 1.445869207, Test RMSE: 0.086180650, Test PSNR: 22.718479156 2022-08-23 10:36:37 - Iter[003000], Epoch[000003], learning rate : 0.000399902, Train Loss: 0.192822084, Test MRAE: 1.584116817, Test RMSE: 0.092025109, Test PSNR: 22.301456451 2022-08-23 10:53:12 - Iter[004000], Epoch[000004], learning rate : 0.000399825, Train Loss: 0.180765674, Test MRAE: 0.614064872, Test RMSE: 0.070524208, Test PSNR: 25.304067612 2022-08-23 11:09:47 - Iter[005000], Epoch[000005], learning rate : 0.000399727, Train Loss: 0.175371751, Test MRAE: 0.542492449, Test RMSE: 0.088396654, Test PSNR: 23.954675674 2022-08-23 11:26:22 - Iter[006000], Epoch[000006], learning rate : 0.000399606, Train Loss: 0.175960690, Test MRAE: 1.136486292, Test RMSE: 0.074130528, Test PSNR: 24.200265884 2022-08-23 11:42:57 - Iter[007000], Epoch[000007], learning rate : 0.000399464, Train Loss: 0.179268226, Test MRAE: 1.133757234, Test RMSE: 0.074117400, Test PSNR: 24.215122223 2022-08-23 11:59:33 - Iter[008000], Epoch[000008], learning rate : 0.000399301, Train Loss: 0.183103830, Test MRAE: 0.690664709, Test RMSE: 0.125816882, Test PSNR: 20.855083466 2022-08-23 12:16:08 - Iter[009000], Epoch[000009], learning rate : 0.000399115, Train Loss: 0.180019125, Test MRAE: 0.625649035, Test RMSE: 0.117592424, Test PSNR: 21.640802383 2022-08-23 12:32:44 - Iter[010000], Epoch[000010], learning rate : 0.000398907, Train Loss: 0.175567344, Test MRAE: 0.705689073, Test RMSE: 0.063903458, Test PSNR: 26.012899399 2022-08-23 12:49:19 - Iter[011000], Epoch[000011], learning rate : 0.000398678, Train Loss: 0.173624143, Test MRAE: 0.719805300, Test RMSE: 0.071538091, Test PSNR: 25.122806549 2022-08-23 13:05:55 - Iter[012000], Epoch[000012], learning rate : 0.000398427, Train Loss: 0.167626292, Test MRAE: 1.066743374, Test RMSE: 0.067167260, Test PSNR: 25.091667175 2022-08-23 13:22:30 - Iter[013000], Epoch[000013], learning rate : 0.000398154, Train Loss: 0.159311280, Test MRAE: 1.099061131, Test RMSE: 0.075757533, Test PSNR: 24.224294662 2022-08-23 13:39:06 - Iter[014000], Epoch[000014], learning rate : 0.000397860, Train Loss: 0.158835545, Test MRAE: 0.538660467, Test RMSE: 0.094124049, Test PSNR: 23.704330444 2022-08-23 13:55:43 - Iter[015000], Epoch[000015], learning rate : 0.000397544, Train Loss: 0.159253776, Test MRAE: 0.668565333, Test RMSE: 0.091043182, Test PSNR: 23.187345505 2022-08-23 15:38:40 - Iter[001000], Epoch[000001], learning rate : 0.000399989, Train Loss: 0.127697811, Test MRAE: 0.745330155, Test RMSE: 0.062896006, Test PSNR: 25.861749649 2022-08-23 15:48:46 - Iter[002000], Epoch[000002], learning rate : 0.000399956, Train Loss: 0.115349077, Test MRAE: 1.359989882, Test RMSE: 0.084344082, Test PSNR: 23.167493820 2022-08-23 16:01:59 - Iter[003000], Epoch[000003], learning rate : 0.000399902, Train Loss: 0.103145719, Test MRAE: 1.378445268, Test RMSE: 0.090944812, Test PSNR: 22.637052536 2022-08-23 16:16:06 - Iter[004000], Epoch[000004], learning rate : 0.000399825, Train Loss: 0.111022204, Test MRAE: 0.556447327, Test RMSE: 0.065509431, Test PSNR: 26.170299530 2022-08-23 16:30:18 - Iter[005000], Epoch[000005], learning rate : 0.000399727, Train Loss: 0.115231283, Test MRAE: 0.518974543, Test RMSE: 0.084985338, Test PSNR: 24.737178802 2022-08-23 16:44:30 - Iter[006000], Epoch[000006], learning rate : 0.000399606, Train Loss: 0.132773608, Test MRAE: 1.072223663, Test RMSE: 0.074294783, Test PSNR: 24.498712540 2022-08-23 16:55:09 - Iter[007000], Epoch[000007], learning rate : 0.000399464, Train Loss: 0.140846550, Test MRAE: 1.058269382, Test RMSE: 0.073174037, Test PSNR: 24.492546082

    What am I doing wrong?

    I would appreciate if you could help me.

    Regards!

    opened by zaidilyas89 6
  • 请问下,我通过cv方法将mat文件转换为的png图片为何是单通道而不是四通道的?

    请问下,我通过cv方法将mat文件转换为的png图片为何是单通道而不是四通道的?

    import h5py import cv2 import numpy as np path = "ARAD_1K_0912.mat" with h5py.File(path, 'r') as mat: hyper = np.float32(np.array(mat['cube']))*255 cv2.imwrite('ARAD_1K_0912.png', hyper[15,:,:]) 作者大佬您好,我通过您提供的代码将预测结果的mat文件转换为了png图片,但是想请问下为何转换后的图片只有单通道?不知我是否理解有误,我在介绍里看着好像转换后的高光谱图像是四通道的?可否解答下我的困惑,感谢您了

    opened by alwaysquietll 4
  • Performance sensitive to Pytorch and CUDA environment.

    Performance sensitive to Pytorch and CUDA environment.

    Hi, thanks for such an interesting work. I observed that the model performs well and good with pytorch 1.8 . However, when I try doing the same in latest pytorch 1.12 and CUDA 11.7, the same model starts overfitting on the training data, and test MRAE does not go down below 0.42. Similarily in pytorch 1.2, the training MRAE starts oscillating around 0.5 and does not change much. Does it mean that the performance of model is highly sensitive to the pytorch or cuda version??

    opened by aksinha340 3
  • 我检测的PSNR与您的不符

    我检测的PSNR与您的不符

    作者您好!针对您的高光谱重建工作,我按照您的方法能跑出啦,也能通过训练的模型进行重建一些图片,可是,在PSNR这个重要验证指标,我测出的数据与您有很大差异,就是,我的PSNR在18-19之间,而您的是34+,所以我就去测试了您上传训练好的MST++模型,经过测试,也是18-19,所以我就不明白我哪里出错了,环境是按照您的环境搭的。我百思不得其解,故通过此渠道寻求您的帮助。如果您能在闲暇之余给予高见,我将感激不尽,谢谢!

    opened by Haitao-Chen-12345 3
  • 请问为什么计算评价指标,只使用图像的中间区域计算呢?

    请问为什么计算评价指标,只使用图像的中间区域计算呢?

    loss_mrae = criterion_mrae(output[:, :, 128:-128, 128:-128], target[:, :, 128:-128, 128:-128]) loss_rmse = criterion_rmse(output[:, :, 128:-128, 128:-128], target[:, :, 128:-128, 128:-128]) loss_psnr = criterion_psnr(output[:, :, 128:-128, 128:-128], target[:, :, 128:-128, 128:-128]) 而不是用 loss_mrae = criterion_mrae(output, target) loss_rmse = criterion_rmse(output, target) loss_psnr = criterion_psnr(output, target)

    opened by yanxinpeng517 2
  • Update README.md

    Update README.md

    maybe there should add :

    test AWAN

    python test.py --data_root ../dataset/ --method awan --pretrained_model_path ./model_zoo/awan.pth --outf ./exp/awan/ --gpu_id 0

    opened by 2016WUJI01 1
Owner
Yuanhao Cai
Yuanhao Cai
Lowest memory consumption and second shortest runtime in NTIRE 2022 challenge on Efficient Super-Resolution

FMEN Lowest memory consumption and second shortest runtime in NTIRE 2022 on Efficient Super-Resolution. Our paper: Fast and Memory-Efficient Network T

null 28 Sep 16, 2022
The 7th edition of NTIRE: New Trends in Image Restoration and Enhancement workshop will be held on June 2022 in conjunction with CVPR 2022.

NTIRE 2022 - Image Inpainting Challenge Important dates 2022.02.01: Release of train data (input and output images) and validation data (only input) 2

Andrés Romero 30 Sep 17, 2022
Tracking code for the winner of track 1 in the MMP-Tracking Challenge at ICCV 2021 Workshop.

Tracking Code for the winner of track1 in MMP-Trakcing challenge This repository contains our tracking code for the Multi-camera Multiple People Track

DamoCV 26 Aug 17, 2022
UDP++ (ECCVW 2020 Oral), (Winner of COCO 2020 Keypoint Challenge).

UDP-Pose This is the pytorch implementation for UDP++, which won the Fisrt place in COCO Keypoint Challenge at ECCV 2020 Workshop. Top-Down Results on

null 20 Jul 29, 2022
Accurate 3D Face Reconstruction with Weakly-Supervised Learning: From Single Image to Image Set (CVPRW 2019). A PyTorch implementation.

Accurate 3D Face Reconstruction with Weakly-Supervised Learning: From Single Image to Image Set —— PyTorch implementation This is an unofficial offici

Sicheng Xu 708 Sep 22, 2022
A Planar RGB-D SLAM which utilizes Manhattan World structure to provide optimal camera pose trajectory while also providing a sparse reconstruction containing points, lines and planes, and a dense surfel-based reconstruction.

ManhattanSLAM Authors: Raza Yunus, Yanyan Li and Federico Tombari ManhattanSLAM is a real-time SLAM library for RGB-D cameras that computes the camera

null 107 Sep 17, 2022
This is the official implementation code repository of Underwater Light Field Retention : Neural Rendering for Underwater Imaging (Accepted by CVPR Workshop2022 NTIRE)

Underwater Light Field Retention : Neural Rendering for Underwater Imaging (UWNR) (Accepted by CVPR Workshop2022 NTIRE) Authors: Tian Ye†, Sixiang Che

jmucsx 11 Aug 31, 2022
DSAC* for Visual Camera Re-Localization (RGB or RGB-D)

DSAC* for Visual Camera Re-Localization (RGB or RGB-D) Introduction Installation Data Structure Supported Datasets 7Scenes 12Scenes Cambridge Landmark

Visual Learning Lab 132 Sep 13, 2022
3DMV jointly combines RGB color and geometric information to perform 3D semantic segmentation of RGB-D scans.

3DMV 3DMV jointly combines RGB color and geometric information to perform 3D semantic segmentation of RGB-D scans. This work is based on our ECCV'18 p

Владислав Молодцов 0 Feb 6, 2022
[CVPRW 2022] Attentions Help CNNs See Better: Attention-based Hybrid Image Quality Assessment Network

Attention Helps CNN See Better: Hybrid Image Quality Assessment Network [CVPRW 2022] Code for Hybrid Image Quality Assessment Network [paper] [code] T

IIGROUP 41 Sep 20, 2022
PyTorch implementation of the Deep SLDA method from our CVPRW-2020 paper "Lifelong Machine Learning with Deep Streaming Linear Discriminant Analysis"

Lifelong Machine Learning with Deep Streaming Linear Discriminant Analysis This is a PyTorch implementation of the Deep Streaming Linear Discriminant

Tyler Hayes 39 Aug 29, 2022
Official PyTorch implementation of "IntegralAction: Pose-driven Feature Integration for Robust Human Action Recognition in Videos", CVPRW 2021

IntegralAction: Pose-driven Feature Integration for Robust Human Action Recognition in Videos Introduction This repo is official PyTorch implementatio

Gyeongsik Moon 29 Sep 24, 2022
[CVPRW 21] "BNN - BN = ? Training Binary Neural Networks without Batch Normalization", Tianlong Chen, Zhenyu Zhang, Xu Ouyang, Zechun Liu, Zhiqiang Shen, Zhangyang Wang

BNN - BN = ? Training Binary Neural Networks without Batch Normalization Codes for this paper BNN - BN = ? Training Binary Neural Networks without Bat

VITA 36 Aug 19, 2022
[CVPRW 2021] Code for Region-Adaptive Deformable Network for Image Quality Assessment

RADN [CVPRW 2021] Code for Region-Adaptive Deformable Network for Image Quality Assessment [Paper on arXiv] Overview Update [2021/5/7] add codes for W

IIGROUP 50 Aug 21, 2022
CVPRW 2021: How to calibrate your event camera

E2Calib: How to Calibrate Your Event Camera This repository contains code that implements video reconstruction from event data for calibration as desc

Robotics and Perception Group 100 Aug 24, 2022
PyTorch version of the paper 'Enhanced Deep Residual Networks for Single Image Super-Resolution' (CVPRW 2017)

About PyTorch 1.2.0 Now the master branch supports PyTorch 1.2.0 by default. Due to the serious version problem (especially torch.utils.data.dataloade

Sanghyun Son 2k Sep 29, 2022
This repo is official PyTorch implementation of MobileHumanPose: Toward real-time 3D human pose estimation in mobile devices(CVPRW 2021).

Github Code of "MobileHumanPose: Toward real-time 3D human pose estimation in mobile devices" Introduction This repo is official PyTorch implementatio

Choi Sang Bum 179 Sep 23, 2022
Pytorch implementation of winner from VQA Chllange Workshop in CVPR'17

2017 VQA Challenge Winner (CVPR'17 Workshop) pytorch implementation of Tips and Tricks for Visual Question Answering: Learnings from the 2017 Challeng

Mark Dong 167 Mar 23, 2022
CoReNet is a technique for joint multi-object 3D reconstruction from a single RGB image.

CoReNet CoReNet is a technique for joint multi-object 3D reconstruction from a single RGB image. It produces coherent reconstructions, where all objec

Google Research 75 Sep 23, 2022