Research map / Detection and segmentation / Open-set and few-shot detection
GKDT: General Keypoint Detection Transformer
Changsheng Lu, Yuxin Chen, Haokun GUI, Rong Wang, Jie Yang, Harry Yang, Anton van den Hengel, Jiaya Jia
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 |
|---|---|---|---|
| tune the DINOv3 visual backbone (ablation) | GKDT without KG transformer · Animal pose | PCK@0.1 visual: 53.86 -> 79.08 | Improved |
| add KG transformer (ablation) | GKDT with frozen backbone · Animal pose | PCK@0.1 visual: 53.86 -> 64.22 | Improved |
| add KG transformer (ablation) | GKDT with tuned backbone · Animal pose | PCK@0.1 visual: 79.08 -> 81.11 | Improved |
| add KG transformer (ablation) | GKDT with frozen backbone · CUB | PCK@0.1 visual: 79.55 -> 77.39 | Hurt |
| add self-attention in KG transformer (ablation) | GKDT · HInt | PCK@0.1 visual: 64.74 -> 69.05 | Improved |
| add cross-attention in KG transformer (ablation) | GKDT · HInt | PCK@0.1 visual: 64.74 -> 65.90 | Improved |
| use dynamic importance sampling instead of uniform sampling (ablation) | OpenKD · HInt | PCK@0.1 both: 47.02 -> 50.94 | Improved |
| use dynamic importance sampling instead of uniform sampling (ablation) | OpenKD · Cephalometric | PCK@0.1 both: 90.99 -> 99.18 | Improved |
| use dynamic importance sampling instead of uniform sampling (ablation) | GKDT · HInt | PCK@0.1 both: 78.18 -> 81.13 | Improved |
| use dynamic importance sampling instead of plain importance sampling (ablation) | GKDT · Cephalometric | PCK@0.1 both: 98.43 -> 99.49 | Improved |
| increase generated kernel size from 1x1 to 5x5 (ablation) | GKDT · Animal pose | PCK@0.1 text: 73.27 -> 70.56 | Hurt |
| increase generated kernel size from 1x1 to 3x3 (ablation) | GKDT · Animal pose | PCK@0.1 text: 73.27 -> 73.89 | No clear effect |
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.
- X-Pose — X-Pose: Detecting Any Keypoints (ECCV 2024) · better on 1 of 1 shared result
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
- OpenKD: Opening Prompt Diversity for Zero- and Few-shot Keypoint Detection (ECCV 2024 · similarity 0.88)
- Exploiting Class-agnostic Visual Prior for Few-shot Keypoint Detection (IJCV 2026 · similarity 0.87)
- Detect Any Keypoints: An Efficient Light-Weight Few-Shot Keypoint Detector (AAAI 2024 · similarity 0.85)
- Generalizable Object Keypoint Localization from Generative Priors (CVPR 2025 · similarity 0.84)
- Back to 3D: Few-Shot 3D Keypoint Detection with Back-Projected 2D Features (CVPR 2024 · similarity 0.84)
- KptLLM: Unveiling the Power of Large Language Model for Keypoint Comprehension (NeurIPS 2024 · similarity 0.83)
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.
