Skip to content

Prachi-Agr/ExtremeDetector

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

3 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

ExtremeDetector

We provide code for low visibility object detection framework - Extreme Detector.

  1. Run framework.py to train CNN to learn degradation present in image.
  2. Run generate_enhanced.py to enhance images based on the degrdation predicted by trained CNN. Accordingly pass the images to image enhancement models.
  3. model_checkpoint contains pre-trained CNN on human action recognition dataset. You can find the dataset here. https://drive.google.com/drive/u/0/folders/1PmxiF1z1UKbDSz1GaA0xsVqkEwjPEFa8
  4. Use the generated enhanced images to train an object detection model. We use YOLOv5 in our experiments.

To generate the low visbility images dataset, we use FoHIS [1] for adding fog effect and Multitask AET with Orthogonal Tangent Regularity for Dark Object Detection [2] for adding low light effect. To generate the snowfall effect, we provide the script in synthetic_snowfall.ipynb

Our enhancement pool consists of pre-trained models for fog removal, low light removal and snowfall removal. We use Zero-DCE [4] for low light image enhancement, FFA-Net [5] for removing fog and Deep Detailed Network [3] for removing snow.

References

[1] @inproceedings{zhang2017towards, title={Towards simulating foggy and hazy images and evaluating their authenticity}, author={Zhang, Ning and Zhang, Lin and Cheng, Zaixi}, booktitle={International Conference on Neural Information Processing}, pages={405--415}, year={2017}, organization={Springer} }.

[2] @InProceedings{Cui_2021_ICCV, author = {Cui, Ziteng and Qi, Guo-Jun and Gu, Lin and You, Shaodi and Zhang, Zenghui and Harada, Tatsuya}, title = {Multitask AET With Orthogonal Tangent Regularity for Dark Object Detection}, booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)}, month = {October}, year = {2021}, pages = {2553-2562} }

[3] @inproceedings{fu2017removing, title={Removing rain from single images via a deep detail network}, author={Fu, Xueyang and Huang, Jiabin and Zeng, Delu and Huang, Yue and Ding, Xinghao and Paisley, John}, booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition}, pages={3855--3863}, year={2017} }

[4] @inproceedings{guo2020zero, title={Zero-reference deep curve estimation for low-light image enhancement}, author={Guo, Chunle and Li, Chongyi and Guo, Jichang and Loy, Chen Change and Hou, Junhui and Kwong, Sam and Cong, Runmin}, booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition}, pages={1780--1789}, year={2020} }

[5] @inproceedings{qin2020ffa, title={FFA-Net: Feature fusion attention network for single image dehazing}, author={Qin, Xu and Wang, Zhilin and Bai, Yuanchao and Xie, Xiaodong and Jia, Huizhu}, booktitle={Proceedings of the AAAI Conference on Artificial Intelligence}, volume={34}, number={07}, pages={11908--11915}, year={2020} }

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published