Research map / Detection and segmentation / CNN features
Camouflaged Instance Segmentation via Explicit De-Camouflaging
Naisong Luo, Yuwen Pan, Rui Sun, Tianzhu Zhang, Zhiwei Xiong, Feng Wu
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 Fourier-amplitude camouflage attribute extraction (ablation) | DCNet · COD10K | AP: 43.1(−2.2) -> 45.3 | Improved |
| add Fourier-amplitude camouflage attribute extraction (ablation) | DCNet · NC4K | AP: 48.1(−4.7) -> 52.8 | Improved |
| use difference attention rather than general subtraction (ablation) | DCNet · COD10K and NC4K | AP: 44.1(−1.2) -> 45.3; 51.3(−1.5) -> 52.8 | Improved |
| add both camouflage attribute extraction and difference attention (ablation) | DCNet · COD10K and NC4K | AP: 40.2(−5.1) -> 45.3; 46.4(−6.4) -> 52.8 | Improved |
| add reference attention instead of cross-attention (ablation) | DCNet · COD10K and NC4K | AP: 42.4 -> 45.3; 49.5 -> 52.8 | Improved |
| add reference attention instead of masked attention (ablation) | DCNet · COD10K and NC4K | AP: 44.7 -> 45.3; 51.7 -> 52.8 | Improved |
| combine multiple feature scales rather than use one scale (ablation) | DCNet · COD10K and NC4K | AP: 45.2 -> 45.3; 52.5 -> 52.8 | Improved |
| use stronger feature extractor backbones (ablation) | DCNet · COD10K and NC4K | AP: 45.3 -> 46.8 -> 50.3 -> 52.3; 52.8 -> 53.5 -> 56.3 -> 58.4 | Improved |
| use DCNet rather than OSFormer with ResNet-50 | COD10K and NC4K · COD10K and NC4K | AP: 41.0 -> 45.3; 42.5 -> 52.8 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- OSFormer — OSFormer: One-Stage Camouflaged Instance Segmentation with Transformers (ECCV 2022) · better on 2 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
- Camouflage Anything: Learning to Hide using Controlled Out-painting and Representation Engineering (CVPR 2025 · similarity 0.87)
- Simultaneously Localize, Segment and Rank the Camouflaged Objects (CVPR 2021 · similarity 0.87)
- VCP-DCN: Beyond Visual Concealed Property via Depth Collaborative Network for Camouflaged Object Detection (ECCV 2026 · similarity 0.87)
- The Making and Breaking of Camouflage (ICCV 2023 · similarity 0.86)
- Shift the Lens: Environment-Aware Unsupervised Camouflaged Object Detection (CVPR 2025 · similarity 0.86)
- CamSAM2: Segment Anything Accurately in Camouflaged Videos (NeurIPS 2025 · similarity 0.85)
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.
