Research map / Detection and segmentation / CNN features
Advanced Discriminative Co-Saliency and Background Mining Transformer for Co-Salient Object Detection
Long Li, Huichao Xie, Nian Liu, Dingwen Zhang, Rao Muhammad Anwer, Hisham Cholakkal, Junwei Han
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 R2R region-to-region inter-image correlation (ablation) | MaskFormer-style baseline · CoCA (ablation) | Sm: 0.6751 -> 0.6945 | Improved |
| add R2R region-to-region inter-image correlation (ablation) | MaskFormer-style baseline · CoCA (ablation) | MAE: 0.1383 -> 0.1234 | Improved |
| add CtP2T contrast-induced token correlation (ablation) | baseline + R2R · CoCA (ablation) | Sm: 0.6945 -> 0.7038 | Improved |
| add CoT2T co-saliency token correlation (ablation) | baseline + R2R + CtP2T · CoCA (ablation) | Sm: 0.7038 -> 0.7140 | Improved |
| add CoT2T co-saliency token correlation (ablation) | baseline + R2R + CtP2T · CoCA (ablation) | MAE: 0.1230 -> 0.1139 | Improved |
| add TGFR token-guided feature refinement (ablation) | DMT without TGFR · CoCA (ablation) | Sm: 0.7140 -> 0.7246 | Improved |
| add TGFR token-guided feature refinement (ablation) | DMT without TGFR · CoCA (ablation) | maxF: 0.6003 -> 0.6190 | Improved |
| add explicit background modeling (BG tokens, CtP2T, BG feature in TGFR) (ablation) | DMT · CoCA (ablation) | Sm: 0.7059 -> 0.7246 | Improved |
| add explicit background modeling (BG tokens, CtP2T, BG feature in TGFR) (ablation) | DMT · CoCA (ablation) | MAE: 0.1259 -> 0.1084 | Improved |
| add distillation step before refusing tokens into features (ablation) | TGFR · CoCA (TGFR variants) | maxF: 0.6046 -> 0.6190 | Improved |
| refuse both co-saliency and BG features instead of co-saliency only (ablation) | TGFR · CoCA (TGFR variants) | Sm: 0.7171 -> 0.7246 | Improved |
| refuse co-saliency first then BG instead of BG first (ablation) | TGFR · CoCA (TGFR variants) | maxF: 0.6092 -> 0.6190 | Improved |
Closest papers by meaning
- Discriminative Co-Saliency and Background Mining Transformer for Co-Salient Object Detection (CVPR 2023 · similarity 0.99)
- Co-Saliency Detection via a Self-Paced Multiple-Instance Learning Framework (TPAMI 2016 · similarity 0.90)
- Detection of Co-salient Objects by Looking Deep and Wide (IJCV 2016 · similarity 0.90)
- Gradient-Induced Co-Saliency Detection (ECCV 2020 · similarity 0.90)
- Self-supervised co-salient object detection via feature correspondences at multiple scales (ECCV 2024 · similarity 0.89)
- Co-Saliency Detection via Mask-Guided Fully Convolutional Networks With Multi-Scale Label Smoothing (CVPR 2019 · similarity 0.89)
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.
