Research map / Architectures and efficiency / CNN design
Learning Dual Convolutional Neural Networks for Low-Level Vision
Jinshan Pan, Sifei Liu, Deqing Sun, Jiawei Zhang, Yang Liu, Jimmy Ren, Zechao Li, Jinhui Tang, Huchuan Lu, Yu-Wing Tai, Ming-Hsuan Yang
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 |
|---|---|---|---|
| jointly estimate structures and details with parallel branches | DualCNN · Super-resolution, Set5/Set14/Urban100/BSDS500 | PSNR | Improved |
| use DualCNN instead of the strongest listed baseline, VDSR | super-resolution · Set5 ×2 | PSNR: 37.6173 -> 37.7005 | Improved |
| use DualCNN instead of the strongest listed baseline, VDSR | super-resolution · Set14 ×4 | PSNR: 28.0667 -> 28.0949 | Improved |
| use DualCNN instead of the strongest listed baseline, VDSR | super-resolution · Urban100 ×4 | PSNR: 25.1575 -> 25.1722 | Improved |
| use DualCNN instead of the strongest listed baseline, VDSR | super-resolution · BSDS500 ×4 | PSNR: 27.2342 -> 27.2454 | Improved |
| use DualCNN instead of VDSR | image filtering · WMF | PSNR: 38.5 -> 39.1 | Improved |
| use DualCNN instead of VDSR | image filtering · RTV | PSNR: 41.6 -> 42.1 | Improved |
| use DualCNN instead of Xu et al. | image filtering · L0 smoothing | PSNR: 32.8 -> 31.4 | Hurt |
| use DualCNN instead of ID-CGAN | deraining · synthetic rainy test dataset | Avg. PSNR: 22.73 -> 24.11 | Improved |
| use DualCNN instead of Ren et al. | dehazing · synthetic hazy image dataset | Avg. PSNR: 18.38 -> 18.85 | Improved |
| add structure and detail branch regularization losses (ablation) | DualCNN dehazing · dehazing validation data | Avg. PSNR: 21.13 -> 26.43 | Improved |
| use structure regularization only (ablation) | DualCNN dehazing · dehazing validation data | Avg. PSNR: 21.13 -> 26.27 | Improved |
Closest papers by meaning
- Dual Convolutional Neural Networks for Low-Level Vision (IJCV 2022 · similarity 0.89)
- DualNet: Learn Complementary Features for Image Recognition (ICCV 2017 · similarity 0.88)
- Towards Understanding the Importance of Shortcut Connections in Residual Networks (NeurIPS 2019 · similarity 0.86)
- Learning Filter Basis for Convolutional Neural Network Compression (ICCV 2019 · similarity 0.85)
- Doubly Convolutional Neural Networks (NeurIPS 2016 · similarity 0.85)
- LSNet: See Large, Focus Small (CVPR 2025 · similarity 0.84)
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.
