Research map / Detection and segmentation / CNN features
Lightweight Pixel Difference Networks for Efficient Visual Representation Learning
Zhuo Su, Jiehua Zhang, Longguang Wang, Hua Zhang, Zhen Liu, Matti Pietikäinen, Li Liu
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 |
|---|---|---|---|
| replace vanilla convolution with PDC in all 16 blocks (ablation) | PiDiNet · BSDS500 val | ODS: 0.772 -> 0.767 | Hurt |
| use sequence of four conv types [CARV]x4 instead of single type repeated (ablation) | PiDiNet · BSDS500 val | ODS: 0.772 -> 0.776 | Improved |
| add PDC in only the first block (ablation) | PiDiNet · BSDS500 val | ODS: 0.772 -> 0.775 | Improved |
| add PDC at multiple scales vs vanilla convolution (ablation) | PiDiNet · BSDS500 val (Basic C=60) | ODS: 0.736 -> 0.757 | Improved |
| add shortcuts (ablation) | PiDiNet · BSDS500 val | ODS: 0.734 -> 0.776 | Improved |
| add CSAM (ablation) | PiDiNet · BSDS500 val | ODS: 0.775 -> 0.776 | Improved |
| add CDCM and CSAM (ablation) | PiDiNet-L · BSDS500 val | ODS: 0.770 -> 0.776 | Improved |
| replace vanilla conv with PDC | Baseline · BSDS500 test | ODS: 0.798 -> 0.807 | Improved |
| replace vanilla conv with PDC | Baseline · BSDS500 test | GPU FPS: 96 -> 92 | Hurt |
| fix binary kernels as LBP patterns instead of learning them (ablation) | Bi-PDC · CIFAR-100 (binary ResNet-18) | top-1 accuracy: 70.49 -> 68.80 | Hurt |
| add Bi-PDC at xi=0.2 channel ratio (ablation) | binary ResNet-18 · CIFAR-100 | top-1 accuracy: performance rises up to xi 0.2 then drops | Mixed |
| replace vanilla binary conv with Bi-CPDC (ablation) | binary ResNet-18 · ImageNet | top-1 accuracy: 60.3 -> 61.6 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- BDCN — BDCN: Bi-Directional Cascade Network for Perceptual Edge Detection (TPAMI 2020) · better on 0 of 3 shared results
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
- Pixel Difference Networks for Efficient Edge Detection (ICCV 2021 · similarity 0.95)
- When Pixel Difference Patterns Meet ViT: PiDiViT for Few-Shot Object Detection (ICCV 2025 · similarity 0.86)
- Rapid Salient Object Detection With Difference Convolutional Neural Networks (TPAMI 2025 · similarity 0.85)
- Bilinear Convolutional Neural Networks for Fine-Grained Visual Recognition (TPAMI 2017 · similarity 0.85)
- Deep Residual Learning for Image Recognition (CVPR 2016 · similarity 0.84)
- 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.
