Research map / Image generation / GANs
ARGAN: Attentive Recurrent Generative Adversarial Network for Shadow Detection and Removal
Bin Ding, Chengjiang Long, Ling Zhang, Chunxia Xiao
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add LSTM layers to the shadow attention detector (ablation) | ARGAN · SBU shadow detection | BER: 7.24 -> 3.09 | Improved |
| add LSTM layers to the shadow attention detector (ablation) | ARGAN · UCF shadow detection | BER: 8.67 -> 3.76 | Improved |
| add LSTM layers to the shadow attention detector (ablation) | ARGAN · ISTD shadow detection | BER: 4.23 -> 2.01 | Improved |
| add LSTM layers to the shadow attention detector (ablation) | ARGAN · SRD shadow removal (shadow regions) | RMSE: 14.68 -> 7.24 | Improved |
| add LSTM layers to the shadow attention detector (ablation) | ARGAN · ISTD shadow removal (whole image) | RMSE: 7.57 -> 6.68 | Improved |
| add semi-supervised training with unlabeled shadow images (ablation) | ARGAN · SBU shadow detection | BER: 3.09 -> 2.56 | Improved |
| add semi-supervised training with unlabeled shadow images (ablation) | ARGAN · ISTD shadow detection | BER: 2.01 -> 1.75 | Improved |
| add semi-supervised training with unlabeled shadow images (ablation) | ARGAN · SRD shadow removal (whole image) | RMSE: 5.74 -> 5.31 | Improved |
| add semi-supervised training with unlabeled shadow images (ablation) | ARGAN · ISTD shadow removal (whole image) | RMSE: 6.68 -> 5.89 | Improved |
| increase progressive steps N from 1 to 3 (ablation) | ARGAN · ISTD shadow detection | BER: 2.22 -> 2.01 | Improved |
| increase progressive steps N from 1 to 3 (ablation) | ARGAN · ISTD shadow removal (whole image) | RMSE: 7.35 -> 6.68 | Improved |
| increase progressive steps N from 3 to 4 (ablation) | ARGAN · ISTD shadow removal (whole image) | RMSE: 6.68 -> 6.67 | No clear effect |
Later papers that compared with this method
1 later accepted paper reports ARGAN as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- Deep Adversarial Decomposition: A Unified Framework for Separating Superimposed Images (CVPR 2020 · better on 1 of 1 shared result)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Stacked Conditional Generative Adversarial Networks for Jointly Learning Shadow Detection and Shadow Removal (CVPR 2018 · similarity 0.93)
- RIS-GAN: Explore Residual and Illumination with Generative Adversarial Networks for Shadow Removal (AAAI 2020 · similarity 0.92)
- Shadow Detection With Conditional Generative Adversarial Networks (ICCV 2017 · similarity 0.90)
- A+D Net: Training a Shadow Detector with Adversarial Shadow Attenuation (ECCV 2018 · similarity 0.90)
- Mask-ShadowGAN: Learning to Remove Shadows From Unpaired Data (ICCV 2019 · similarity 0.90)
- Towards Ghost-Free Shadow Removal via Dual Hierarchical Aggregation Network and Shadow Matting GAN (AAAI 2020 · similarity 0.90)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
