CTRL-C: Camera calibration TRansformer with Line-Classification
This repository contains the official code and pretrained models for CTRL-C (Camera calibration TRansformer with Line-Classification). Jinwoo Lee, Hyunsung Go, Hyunjoon Lee, Sunghyun Cho, Minhyuk Sung and Junho Kim. ICCV 2021.
Single image camera calibration is the task of estimating the camera parameters from a single input image, such as the vanishing points, focal length, and horizon line. In this work, we propose Camera calibration TRansformer with Line-Classification (CTRL-C), an end-to-end neural network-based approach to single image camera calibration, which directly estimates the camera parameters from an image and a set of line segments. Our network adopts the transformer architecture to capture the global structure of an image with multi-modal inputs in an end-to-end manner. We also propose an auxiliary task of line classification to train the network to extract the global geometric information from lines effectively. Our experiments demonstrate that CTRL-C outperforms the previous state-of-the-art methods on the Google Street View and SUN360 benchmark datasets.
Results & Checkpoints
Dataset | Up Dir (◦) | Pitch (◦) | Roll (◦) | FoV (◦) | AUC (%) | URL |
---|---|---|---|---|---|---|
Google Street View | 1.80 | 1.58 | 0.66 | 3.59 | 87.29 | gdrive |
SUN360 | 1.91 | 1.50 | 0.96 | 3.80 | 85.45 | gdrive |
Preparation
-
Clone this repository
-
Setup environments
conda create -n ctrlc python conda activate ctrlc conda install -c pytorch torchvision pip install -r requrements.txt
Training Datasets
- Google Street View dataset
- SUN360 dataset
- You need to preprocess the dataset
Training
- Single GPU
python main.py --config-file 'config-files/ctrl-c.yaml' --opts OUTPUT_DIR 'logs'
- Multi GPU
python -m torch.distributed.launch --nproc_per_node=4 --use_env main.py --config-file 'config-files/ctrl-c.yaml' --opts OUTPUT_DIR 'logs'
Evaluation
python test.py --dataset 'GoogleStreetView' --opts OUTPUT_DIR 'outputs'
Citation
If you use this code for your research, please cite our paper:
@InProceedings{Lee:2021:ICCV,
Title = {{CTRL-C: Camera calibration TRansformer with Line-Classification}},
Author = {Jinwoo Lee and Hyunsung Go and Hyunjoon Lee and Sunghyun Cho and Minhyuk Sung and Junho Kim},
Booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)},
Year = {2021},
}
License
CTRL-C is released under the Apache 2.0 license. Please see the LICENSE file for more information.
Acknowledgments
This code is based on the implementations of DETR: End-to-End Object Detection with Transformers.