Research map / Detection and segmentation / Weakly supervised detection and segmentation
Probabilistic Warp Consistency for Weakly-Supervised Semantic Correspondences
Prune Truong, Martin Danelljan, Fisher Yu, Luc Van Gool
Check 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 |
|---|---|---|---|
| Use probabilistic warp consistency weak supervision | SF-Net · PF-Pascal | PCK at αimg=0.10: 84.0 -> 87.6 | Improved |
| Use probabilistic warp consistency weak supervision | SF-Net · PF-Willow | PCK at αbbox-kp=0.10: 74.0 -> 78.3 | Improved |
| Use probabilistic warp consistency weak supervision | NC-Net · SPair-71K | PCK at αbbox=0.10: 28.8 -> 35.3 | Improved |
| Add probabilistic losses to keypoint supervision | NC-Net* · TSS | PCK at αimg=0.05: 75.3 -> 91.2 | Improved |
| Add a visibility mask to PW-bipath training (ablation) | SF-Net · PF-Pascal | PCK at αimg=0.10: 82.3 -> 83.7 | Improved |
| Add PWarp-supervision to visibility-aware PW-bipath (ablation) | SF-Net · PF-Pascal | PCK at αimg=0.10: 83.7 -> 84.9 | Improved |
| Add PNeg unmatched-state supervision (ablation) | SF-Net · PF-Pascal | PCK at αimg=0.10: 84.9 -> 87.6 | Improved |
| Replace mapping-based Warp Consistency with probabilistic warp consistency (ablation) | SF-Net · SPair-71K | PCK at αbbox=0.10: 26.6 -> 33.5 | Improved |
| Use only PWarp-supervision instead of the full objective (ablation) | SF-Net · PF-Pascal | PCK at αimg=0.10: 87.6 -> 74.3 | Hurt |
| Add probabilistic PW-bipath to keypoint supervision (ablation) | SF-Net* · PF-Willow | PCK at αbbox-kp=0.10: 72.5 -> 77.9 | Improved |
| Add PWarp-supervision to PW-bipath and keypoint supervision (ablation) | SF-Net* · SPair-71K | PCK at αbbox=0.10: 31.1 -> 32.5 | Improved |
| Add PWarp-supervision (ablation) | PWarpC-NC-Net · Semantic matching benchmarks | PCK | Hurt |
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.
- SF-Net — SFNet: Learning Object-Aware Semantic Correspondence (CVPR 2019) · better on 5 of 5 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
- Warp Consistency for Unsupervised Learning of Dense Correspondences (ICCV 2021 · similarity 0.91)
- Weakly Supervised Learning of Semantic Correspondence through Cascaded Online Correspondence Refinement (ICCV 2023 · similarity 0.88)
- WarpNet: Weakly Supervised Matching for Single-View Reconstruction (CVPR 2016 · similarity 0.88)
- End-to-End Weakly-Supervised Semantic Alignment (CVPR 2018 · similarity 0.88)
- Show, Match and Segment: Joint Weakly Supervised Learning of Semantic Matching and Object Co-Segmentation (TPAMI 2020 · similarity 0.87)
- SD4Match: Learning to Prompt Stable Diffusion Model for Semantic Matching (CVPR 2024 · 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.
