donggong1 / learn-optimizer-rgdn

Learning Deep Gradient Descent Optimization for Image Deconvolution. Dong Gong, Zhen Zhang, Qinfeng Shi, Anton van den Hengel, Chunhua Shen, Yanning Zhang. In IEEE Transactions on Neural Networks and Learning Systems (TNNLS), 2020.

https://arxiv.org/pdf/1804.03368.pdf

MIT License

32 stars 11 forks source link

blur-kernel-generation deep-learning image-deconvolution optimizer

readme

learn-optimizer-rgdn

Learning Deep Gradient Descent Optimization for Image Deconvolution

Dong Gong, Zhen Zhang, Qinfeng Shi, Anton van den Hengel, Chunhua Shen, Yanning Zhang (*Equal contribution)

In IEEE Transactions on Neural Networks and Learning Systems (TNNLS), 2020.

This implementation is for non-commercial research use only. If you find this code useful in your research, please cite the above paper.

Testing script: ./test_entry.py.
Pretrained model: ./rgdn.tr.
Testing Examples, with real and synthetic images.
Kernel generation in Python: ./data/make_kernel.py, a Python implementation of the kernel simulation method from "A. Chakrabarti, A neural approach to blind motion deblurring, in ECCV, 2016".

Requirements

Python 3
PyTorch 1.1.0 (recommended version)