vita-group / dehaze

[Preprint] "Improved Techniques for Learning to Dehaze and Beyond: A Collective Study"

Improved Techniques for Learning to Dehaze and Beyond: A Collective Study

Introduction

This is the official codebase for our paper "Improved Techniques for Learning to Dehaze and Beyond: A Collective Study".

The paper reviews the collective endeavors by the team of authors in exploring two interlinked important tasks, based on the recently released REalistic Single Image DEhazing (RESIDE) benchmark: i) single image dehazing as a low-level image restoration problem; ii) high-level visual understanding (e.g., object detection) from hazy images. For the first task, the authors investigated on a variety of loss functions, and found perception-driven loss to improve dehazing performance very notably. For the second task, the authors came up with multiple solutions including using more advanced modules in the dehazing-detection cascade, as well as domain-adaptive object detectors. In both tasks, our proposed solutions are verified to significantly advance the state-of-the-art performance.

Code organization

Each individual software package and corresponding documentation are located under code/PACKAGE_NAME

PAD-Net

See code/pad_net

Domain adaptation for MaskRNN

See code/adapt_maskrnn

Improving Object Detection in Haze

See code/iodh

Sandeep and Satya's work

see code/sandeep_satya

Acknowledgements

This collective study was initially performed as a team project effort in the Machine Learning course (CSCE 633, Spring 2018) of CSE@TAMU, taught by Dr. Zhangyang Wang. We acknowledge the Texas A&M High Performance Research Computing (HPRC) for providing a part of the computing resources used in this research.

Contact

• Yu Liu: [email protected]
• Guanlong Zhao: [email protected]
• Boyuan Gong
• Yang Li
• Ritu Raj
• Niraj Goel
• Satya Kesav
• Sandeep Gottimukkala
• Zhangyang Wang: [email protected]
• Wenqi Ren
• Dacheng Tao

Citation

@article{liu2018dehaze,
  title={Improved Techniques for Learning to Dehaze and Beyond: A Collective Studys},
  author={Yu Liu, Guanlong Zhao, Boyuan Gong, Yang Li, Ritu Raj, Niraj Goel, Satya Kesav, Sandeep Gottimukkala, Zhangyang Wang, Wenqi Ren, Dacheng Tao},
  journal={TBD},
  year={2018}
}

dehaze's Issues

A Problem about "StopIteration"

When I run the "run_me.py",there is a problem as follows:
Using TensorFlow backend.
Traceback (most recent call last):
File "F:/dehaze/code/adapt_maskcnn/codes/run_me.py", line 23, in
test_image_names = next(os.walk(test_image_dir))[2]
StopIteration

Could you please give me a solution?Thanks.

DMASK RCNN2 pretrained model not accessible

Howdy!

The link for Dmask-RCNN2 pretrained model given under README file for adapt-rcnn is not accessible. Could you please provide a new link?

Thanks