Research map / Detection and segmentation / CNN features
Feature Shrinkage Pyramid for Camouflaged Object Detection With Transformers
Read the paperCheck 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 |
|---|---|---|---|
| Use FSPNet with locality-enhanced ViT features and progressive feature shrinkage | ViT-based camouflaged object detector · CAMO | S-measure: .820 -> .856 | Helped |
| Use FSPNet with locality-enhanced ViT features and progressive feature shrinkage | ViT-based camouflaged object detector · COD10K | weighted F-measure: .729 -> .735 | Helped |
| Use FSPNet with locality-enhanced ViT features and progressive feature shrinkage | ViT-based camouflaged object detector · NC4K | S-measure: .853 -> .879 | Helped |
| Aggregate all transformer feature layers instead of fewer layers (ablation) | ViT baseline decoder · CAMO | S-measure: .774 -> .798 | Helped |
| Add FSD to the B12 backbone features (ablation) | ViT backbone · CAMO | S-measure: .798 -> .817 | Helped |
| Add NL-TEM to B12+D (ablation) | FSPNet without NL-TEM · CAMO | S-measure: .817 -> .856 | Helped |
| Replace FSD with a U-shaped decoder (ablation) | FSPNet · COD10K | S-measure: .851 -> .825 | Hurt |
| Replace AIM with concatenation and 1×1 convolution (ablation) | FSPNet · COD10K | S-measure: .851 -> .848 | Hurt |
| Extend AIM to aggregate three adjacent feature layers (ablation) | FSPNet · COD10K | S-measure: .851 -> .840 | Hurt |
| Use overlapping pairwise feature aggregation and remove lateral supervision and within-layer feature interaction (ablation) | FSPNet · COD10K | S-measure: .849 -> .851 | Hurt |
| Add lateral supervision and within-layer feature interaction to the overlapping decoder (ablation) | overlapping pairwise decoder · COD10K | S-measure: .849 -> .850 | Helped |
| Add GFM to the B12+D model (ablation) | B12+D · COD10K | S-measure: .844 -> .846 | Helped |
Closest papers by meaning
- Pyramidal Feature Shrinking for Salient Object Detection (AAAI 2021 · similarity 0.89)
- RelationNet++: Bridging Visual Representations for Object Detection via Transformer Decoder (NeurIPS 2020 · similarity 0.88)
- Zoom in and Out: A Mixed-Scale Triplet Network for Camouflaged Object Detection (CVPR 2022 · similarity 0.88)
- SDDF: Specificity-Driven Dynamic Focusing for Open-Vocabulary Camouflaged Object Detection (CVPR 2026 · similarity 0.87)
- TGBFormer: Transformer-GraphFormer Blender Network for Video Object Detection (AAAI 2025 · similarity 0.87)
- SdalsNet: Self-Distilled Attention Localization and Shift Network for Unsupervised Camouflaged Object Detection (AAAI 2025 · similarity 0.86)
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.