Research map / Detection and segmentation / Open-set and few-shot detection
The Functional Correspondence Problem
Zihang Lai, Senthil Purushwalkam, Abhinav Gupta
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 task-embedding-conditioned gating (vs uniform gating) (ablation) | modular network · FunKPoint functional correspondence | PCK: 52.8 -> 58.4 | Improved |
| add task-embedding-conditioned gating (vs single learned gating) (ablation) | modular network · FunKPoint functional correspondence | PCK: 52.5 -> 58.4 | Improved |
| train on FunKPoint with modular network vs ImageNet features | ResNet50 conv4_x · FunKPoint functional correspondence | PCK: 22.0 -> 58.4 | Improved |
| initialize trunk from ImageNet instead of DINO (ablation) | task-driven model · FunKPoint functional correspondence | PCK: 43.5 -> 58.4 | Improved |
| DINO-initialized model vs ImageNet baseline | task-driven model · FunKPoint functional correspondence | PCK: 22.0 -> 43.5 | Improved |
| use functional correspondence representation vs ImageNet | linear probe · Few-shot classification, 1-shot | accuracy: 44.68 -> 47.46 | Improved |
| use functional correspondence representation vs ImageNet | linear probe · Few-shot classification, 2-shot | accuracy: 52.52 -> 55.68 | Improved |
| use functional correspondence representation vs ImageNet | linear probe · Few-shot classification, 5-shot | accuracy: 54.63 -> 56.32 | Improved |
| use functional correspondence vs finetuning on FunKPoint object classification | linear probe · Few-shot classification, 1-shot | accuracy: 45.03 -> 47.46 | Improved |
| use functional representation vs ImageNet | grasp network · Grasp dataset | accuracy: 88.17 -> 89.85 | Improved |
| use functional representation vs ImageNet finetuned on FunKPoint | grasp network · Grasp dataset | accuracy: 88.64 -> 89.85 | Improved |
Closest papers by meaning
- Weakly-Supervised Learning of Dense Functional Correspondences (ICCV 2025 · similarity 0.90)
- Zero-Shot Image Feature Consensus with Deep Functional Maps (ECCV 2024 · similarity 0.88)
- Semantic Correspondence: Unified Benchmarking and a Strong Baseline (TPAMI 2025 · similarity 0.88)
- Telling Left from Right: Identifying Geometry-Aware Semantic Correspondence (CVPR 2024 · similarity 0.87)
- Jamais Vu: Exposing the Generalization Gap in Supervised Semantic Correspondence (NeurIPS 2025 · similarity 0.87)
- Beyond the Doors of Perception: Vision Transformers Represent Relations Between Objects (NeurIPS 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.
