SenseEarth2020 - ChangeDetection
1st place in the Satellite Image Change Detection Challenge hosted by SenseTime.
Our Method
Task Description
Given two images of the same scene acquired at different times, we are required to mark the changed and unchanged areas. Moreover, as for the changed areas, we need to annotate their detailed semantic masks.
The change detection task in this competition can be decomposed into two sub-tasks:
- binary segmentation of changed and unchanged areas.
- semantic segmentation of changed areas.
Model
Pseudo Labeling
The core practice is using self-distillation strategy to assign pseudo labels to unchanged areas.
Specifically, in our experiments, predictions of five HRNet-based segmentation models are ensembled, serving as pseudo labels of unchanged areas.
The overall training process can be summarized as:
- Training multiple large segmentation models.
- Ensembling their predictions on unchanged areas.
- Training a smaller model with both labeled and pseudo labeled areas.
For more details, please refer to the technical report and presentation.
Getting Started
Dataset
Description | Download [password: f3qq]
Pretrained Model
HRNet-W18 | HRNet-W40 | HRNet-W44 | HRNet-W48 | HRNet-W64
Final Trained Model
PSPNet-HRNet-W18 | PSPNet-HRNet-W40
File Organization
# store the whole dataset and pretrained backbones
mkdir -p data/dataset ; mkdir -p data/pretrained_models ;
# store the trained models
mkdir -p outdir/models ;
# store the pseudo masks
mkdir -p outdir/masks/train/im1 ; mkdir -p outdir/masks/train/im2 ;
# store predictions of validation set and testing set
mkdir -p outdir/masks/val/im1 ; mkdir -p outdir/masks/val/im2 ;
mkdir -p outdir/masks/test/im1 ; mkdir -p outdir/masks/test/im2 ;
├── data
├── dataset # download from the link above
│ ├── train # training set
| | ├── im1
| | └── ...
│ └── val # the final testing set (without labels)
|
└── pretrained_models
├── hrnet_w18.pth
├── hrnet_w40.pth
└── ...
Training
# Please refer to utils/options.py for more arguments
# If hardware supports, more backbones can be trained, such as hrnet_w44, hrnet_w48
CUDA_VISIBLE_DEVICES=0,1,2,3 python train.py --backbone hrnet_w18 --pretrained --model pspnet --lightweight
Pseudo Labeling & Re-training
# This step is optional but important in performance improvement
# Modify the backbones, models and checkpoint paths in L20-40 in label.py manually according to your saved models
# It is better to ensemble multiple trained models for pseudo labeling
# Pseudo labeling
CUDA_VISIBLE_DEVICES=0,1,2,3 python label.py
# Re-training
CUDA_VISIBLE_DEVICES=0,1,2,3 python train.py --backbone hrnet_w18 --pretrained --model pspnet --lightweight --use-pseudo-label
Testing
# Modify the backbones, models and checkpoint paths in L39-44 in test.py manually according to your saved models
# Or simply use our final trained models
CUDA_VISIBLE_DEVICES=0,1,2,3 python test.py
22 Nov 30, 2022
12 Oct 25, 2022
4 Jul 1, 2022
23 Mar 9, 2022
46 Dec 2, 2022
37 Oct 30, 2022
2 Jan 17, 2022
5 Nov 9, 2022
91 Dec 21, 2022
30 Nov 22, 2022
24 Nov 23, 2022
79 Oct 8, 2022
18 Aug 19, 2022
82 Dec 29, 2022
9 Dec 1, 2022
94 Dec 25, 2022
158 Jan 8, 2023
11 Mar 19, 2022