Research map / Detection and segmentation / CNN features
Discriminative Co-Saliency and Background Mining Transformer for Co-Salient Object Detection
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 DMT instead of the strongest listed baseline for each metric | CoSOD models · CoCA | S_m: 0.7101 -> 0.7246 | Helped |
| Use DMT instead of the strongest listed baseline for each metric | CoSOD models · CoSal2015 | S_m: 0.8666 -> 0.8974 | Helped |
| Use DMT instead of the strongest listed baseline for each metric | CoSOD models · CoSOD3k | S_m: 0.8191 -> 0.8514 | Helped |
| Add R2R inter-image region correlation (ablation) | MaskFormer-style baseline · CoCA | S_m: 0.6751 -> 0.6945 | Helped |
| Replace vanilla P2T with contrast-induced CtP2T (ablation) | R2R CoSOD model · CoCA | S_m: 0.6945 -> 0.7038 | Helped |
| Add CoT2T group consensus modeling (ablation) | R2R and CtP2T model · CoCA | S_m: 0.7038 -> 0.7140 | Helped |
| Add token-guided feature refinement (ablation) | R2R, CtP2T, and CoT2T model · CoCA | S_m: 0.7140 -> 0.7246 | Helped |
| Remove explicit BG modeling (ablation) | DMT · CoCA | S_m: 0.7246 -> 0.7059 | Hurt |
| Remove distillation and fuse tokens directly with features (ablation) | TGFR · CoCA | S_m: 0.7140 -> 0.7141 | No clear effect |
| Refuse only the co-saliency distilled feature (ablation) | TGFR · CoCA | S_m: 0.7140 -> 0.7171 | Helped |
| Refuse only the BG distilled feature (ablation) | TGFR · CoCA | S_m: 0.7140 -> 0.7155 | Helped |
| Refuse both distilled features, BG first then co-saliency (ablation) | TGFR · CoCA | S_m: 0.7140 -> 0.7197 | Helped |
Closest papers by meaning
- Co-Saliency Detection via Mask-Guided Fully Convolutional Networks With Multi-Scale Label Smoothing (CVPR 2019 · similarity 0.89)
- Democracy Does Matter: Comprehensive Feature Mining for Co-Salient Object Detection (CVPR 2022 · similarity 0.88)
- TF-SSD: A Strong Pipeline via Synergic Mask Filter for Training-free Co-salient Object Detection (CVPR 2026 · similarity 0.88)
- Co-Saliency Detection Within a Single Image (AAAI 2018 · similarity 0.88)
- CoADNet: Collaborative Aggregation-and-Distribution Networks for Co-Salient Object Detection (NeurIPS 2020 · similarity 0.87)
- Unsupervised Salient Instance Detection (CVPR 2024 · 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.