Research map / Detection and segmentation / Open-set and few-shot detection
Semantic-Aware Fine-Grained Correspondence
Yingdong Hu, Renhao Wang, Kaifeng Zhang, Yang Gao
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 |
|---|---|---|---|
| fuse semantic MoCo features with fine-grained FC features | FC · DAVIS-2017 val | J&F_m: 67.7 -> 71.2 | Improved |
| add fine-grained FC features to MoCo | MoCo · DAVIS-2017 val | J&F_m: 65.9 -> 71.2 | Improved |
| add BYOL image-level loss to FC in joint training (ablation) | FC · DAVIS-2017 val | J&F_m: 67.7 -> 67.2 | Hurt |
| add color jittering to cropping (ablation) | FC · DAVIS-2017 val | J&F_m: 67.6 -> 65.5 | Hurt |
| add color dropping (grayscale) to cropping (ablation) | FC · DAVIS-2017 val | J&F_m: 67.6 -> 61.9 | Hurt |
| add Gaussian blur to cropping (ablation) | FC · DAVIS-2017 val | J&F_m: 67.6 -> 65.9 | Hurt |
| add random flip to cropping (ablation) | FC · DAVIS-2017 val | J&F_m: 67.6 -> 67.7 | No clear effect |
| increase feature map resolution from 8x8 to 32x32 (ablation) | FC · DAVIS-2017 val | J&F_m: 63.3 -> 67.6 | Improved |
| fuse two networks of the same correspondence type (BYOL + MoCo) (ablation) | MoCo · DAVIS-2017 val | J&F_m: 65.9 -> 67.5 | No clear effect |
| fuse UVC with FC (ablation) | FC · DAVIS-2017 val | J&F_m: 67.7 -> 61.5 | Hurt |
| pretrain FC on COCO instead of YouTube-VOS (ablation) | FC · DAVIS-2017 val | J&F_m: 67.7 -> 68.2 | No clear effect |
| pretrain FC on ImageNet instead of YouTube-VOS (ablation) | FC · DAVIS-2017 val | J&F_m: 67.7 -> 67.9 | No clear effect |
Closest papers by meaning
- Joint-task Self-supervised Learning for Temporal Correspondence (NeurIPS 2019 · similarity 0.92)
- Rethinking Self-Supervised Correspondence Learning: A Video Frame-Level Similarity Perspective (ICCV 2021 · similarity 0.90)
- Learning Fine-Grained Features for Pixel-Wise Video Correspondences (ICCV 2023 · similarity 0.90)
- Semantic Correspondence: Unified Benchmarking and a Strong Baseline (TPAMI 2025 · similarity 0.90)
- Learning Contrastive Representation for Semantic Correspondence (IJCV 2022 · similarity 0.89)
- Self-Supervised Cross-View Correspondence with Predictive Cycle Consistency (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.
