Residual-Conditioned Optimal Transport: Towards Structure-Preserving Unpaired and Paired Image Restoration
Authors: Xiaole Tang, Xin Hu, Xiang Gu, Jian Sun
ICML 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive experiments on multiple restoration tasks show that RCOT achieves competitive performance in terms of both distortion measures and perceptual quality, restoring images with more faithful structures as compared with state-of-the-art methods. |
| Researcher Affiliation | Academia | 1School of Mathematics and Statistics, Xi an Jiaotong University, Xi an, China. Correspondence to: Jian Sun <jiansun@xjtu.edu.cn>. |
| Pseudocode | Yes | Algorithm 1 RCOT Solver to compute the OT map. |
| Open Source Code | No | The source code will be released after the possible publication of our work. |
| Open Datasets | Yes | For Gaussian image denoising, we train the model on a combination of BSD400 (Arbelaez et al., 2010) and WED (Ma et al., 2016) datasets. ...We evaluate RCOT on the Kodak24 (Franzen, 1999) and CBSD68 (Martin et al., 2001) datasets... For Rain100L (Yang et al., 2017)... For real-world SPANet (Wang et al., 2019)... We evaluate RCOT on the synthetic SOTS (Li et al., 2018a) dataset... and real hazy O-HAZE (Ancuti et al., 2018) dataset... For image super-resolution, we evaluate the performance on the challenging DIV2K (Agustsson & Timofte, 2017) dataset... |
| Dataset Splits | Yes | For Rain100L (Yang et al., 2017), we train the model with 13,712 paired clean-rain images... For real-world SPANet (Wang et al., 2019), it contains 27.5K paired rainy and rain-free images for training, and 1, 000 paired images for testing. ...SOTS (Li et al., 2018a) dataset, which contains 72,135 images for training and 500 images for testing... O-HAZE... 40 images are used for training and the other 5 images are used for testing. ...DIV2K... 800 (4x) LR and HR image pairs for training and 100 pairs for testing. |
| Hardware Specification | Yes | All the experiments are conducted on the Pytorch framework with an NVIDIA 4090 GPU. |
| Software Dependencies | No | All the experiments are conducted on the Pytorch framework with an NVIDIA 4090 GPU. (Only mentions PyTorch without a version, and no other specific software dependencies with versions). |
| Experiment Setup | Yes | We train separate models for different tasks using the RMSProp optimizer with a learning rate of 1 10 4 for the transport network Tθ and 0.5 10 4 for the potential network φw. The inner iteration number n T is set to be 1. The learning rate is decayed by a factor of 10 after 100 epochs. ...During training, we crop patches of size 256x256 as input and use a batch size of 4. |