Research map / Architectures and efficiency / CNN design
ADCrowdNet: An Attention-Injective Deformable Convolutional Network for Crowd Understanding
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 soft AMG attention (ablation) | DME · ShanghaiTech Part_A | MAE: 68.5 -> 66.1 | Helped |
| add soft AMG attention (ablation) | DME · ShanghaiTech Part_B | MAE: 9.3 -> 7.6 | Helped |
| add soft AMG attention (ablation) | DME · WorldExpo'10 | MAE: 8.5 -> 7.4 | Helped |
| add soft AMG attention (ablation) | DME · UCSD | MAE: 0.98 -> 1.10 | Hurt |
| add soft AMG attention (ablation) | DME · UCF_CC_50 | MAE: 257.1 -> 257.9 | No clear effect |
| use binary attention instead of soft attention (ablation) | ADCrowdNet (AMG-DME) · ShanghaiTech Part_A | MAE: 66.1 -> 63.2 | Helped |
| use binary attention instead of soft attention (ablation) | ADCrowdNet (AMG-DME) · ShanghaiTech Part_B | MAE: 7.6 -> 8.2 | Hurt |
| use binary attention instead of soft attention (ablation) | ADCrowdNet (AMG-DME) · UCF_CC_50 | MAE: 257.9 -> 266.4 | Hurt |
| inject attention into DME front-end features rather than weighting the input image (ablation) | ADCrowdNet · ShanghaiTech Part_A | MAE: 66.1 -> 70.9 | Hurt |
| inject attention into DME front-end features rather than weighting the input image (ablation) | ADCrowdNet · WorldExpo'10 | MAE: 7.4 -> 7.3 | Helped |
| set binary-attention threshold to 0.1 (ablation) | ADCrowdNet (AMG-bAttn-DME) · ShanghaiTech Part_A | MAE: 68.0 -> 63.2 | Helped |
| set binary-attention threshold to 0.1 (ablation) | ADCrowdNet (AMG-bAttn-DME) · ShanghaiTech Part_B | MAE: 9.2 -> 8.2 | Helped |
Closest papers by meaning
- Attention Scaling for Crowd Counting (CVPR 2020 · similarity 0.89)
- Shallow Feature Based Dense Attention Network for Crowd Counting (AAAI 2020 · similarity 0.88)
- Single-Image Crowd Counting via Multi-Column Convolutional Neural Network (CVPR 2016 · similarity 0.87)
- Crowd Counting and Density Estimation by Trellis Encoder-Decoder Networks (CVPR 2019 · similarity 0.87)
- Gramformer: Learning Crowd Counting via Graph-Modulated Transformer (AAAI 2024 · similarity 0.85)
- CrowdDiff: Multi-hypothesis Crowd Density Estimation using Diffusion Models (CVPR 2024 · 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.