Research map / Detection and segmentation / CNN features
When W4A4 Breaks Camouflaged Object Detection: Token-Group Dual-Constraint Activation Quantization
Tianqi Li, Wenyu Fang, Xin He, Xue Geng, Xu Cheng, Yun 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 |
|---|---|---|---|
| quantize activations to 4 bits (W4A8 to W4A4) | CFRN · NC4K | S-alpha: .882 -> .443 | Hurt |
| quantize weights to 4 bits at 8-bit activations (W8A8 to W4A8) | CFRN · NC4K | S-alpha: .887 -> .882 | No clear effect |
| add DSTG and DCRP together to naive W4A4 (ablation) | CFRN · NC4K | S-alpha: .443 -> .837 | Improved |
| add DSTG only (without DCRP) (ablation) | CFRN · NC4K | S-alpha: .443 -> .451 | No clear effect |
| add DCRP only (without DSTG) (ablation) | CFRN · NC4K | S-alpha: .443 -> .435 | Hurt |
| add DCRP only (without DSTG) (ablation) | ESCNet · NC4K | S-alpha: .576 -> .771 | Improved |
| add DSTG on top of DCRP (ablation) | ESCNet · NC4K | S-alpha: .771 -> .881 | Improved |
| use per-tensor ranges instead of token groups (ablation) | CFRN · NC4K | S-alpha: .443 -> .407 | Hurt |
| use COD-TDQ instead of best PTQ baseline RepQ-ViT | CFRN · NC4K | S-alpha: .718 -> .837 | Improved |
| deploy COD-TDQ W4A4 with INT4 path instead of FP32 | CFRN · runtime | FPS (img/s): 29.40 -> 44.21 | Improved |
Closest papers by meaning
- SAM-COD: SAM-guided Unified Framework for Weakly-Supervised Camouflaged Object Detection (ECCV 2024 · similarity 0.88)
- CODiff: One-Step Diffusion Model for Camouflaged Object Detection (ICML 2026 · similarity 0.88)
- Beyond Weak Supervision: MLLMs-Guided Graded Knowledge Distillation for Unsupervised Camouflaged Object Detection (CVPR 2026 · similarity 0.88)
- Just a Hint: Point-Supervised Camouflaged Object Detection (ECCV 2024 · similarity 0.88)
- Feature Shrinkage Pyramid for Camouflaged Object Detection With Transformers (CVPR 2023 · similarity 0.88)
- Shift the Lens: Environment-Aware Unsupervised Camouflaged Object Detection (CVPR 2025 · similarity 0.87)
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.
