Papers › Keypoint Transformer: Solving Joint Identification in Challenging Hands and Object...

Keypoint Transformer: Solving Joint Identification in Challenging Hands and Object Interactions for Accurate 3D Pose Estimation

29 Apr 2021CVPR 2022 1arXiv:2104.14639archive 2025-07-28

Shreyas Hampali, Sayan Deb Sarkar, Mahdi Rad, Vincent Lepetit

We propose a robust and accurate method for estimating the 3D poses of two hands in close interaction from a single color image. This is a very challenging problem, as large occlusions and many confusions between the joints may happen. State-of-the-art methods solve this problem by regressing a heatmap for each joint, which requires solving two problems simultaneously: localizing the joints and recognizing them. In this work, we propose to separate these tasks by relying on a CNN to first localize joints as 2D keypoints, and on self-attention between the CNN features at these keypoints to associate them with the corresponding hand joint. The resulting architecture, which we call "Keypoint Transformer", is highly efficient as it achieves state-of-the-art performance with roughly half the number of model parameters on the InterHand2.6M dataset. We also show it can be easily extended to estimate the 3D pose of an object manipulated by one or two hands with high performance. Moreover, we created a new dataset of more than 75,000 images of two hands manipulating an object fully annotated in 3D and will make it publicly available.

PaperPDFConference PDFCode

In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

Code

shreyashampali/kypt_transformer officialmentioned on GitHubpytorch report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

3D Hand Pose Estimation3D Interacting Hand Pose Estimation3D Pose EstimationPose Estimationhand-object pose

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
3D Hand Pose Estimation HO-3D v2 KPT-Transformer AUC_J 0.786 #18 of 24 Archive leaderboard report
3D Hand Pose Estimation HO-3D v2 KPT-Transformer PA-MPJPE (mm) 10.8 #18 of 24 Archive leaderboard report
3D Hand Pose Estimation HO-3D v3 KPT-Transformer AUC_J 0.785 #7 of 8 Archive leaderboard report
3D Hand Pose Estimation HO-3D v3 KPT-Transformer PA-MPJPE 10.9 #7 of 8 Archive leaderboard report
3D Interacting Hand Pose Estimation InterHand2.6M Keypoint Transformer MPJPE Test 12.78 #6 of 9 Archive leaderboard report
3D Interacting Hand Pose Estimation InterHand2.6M Keypoint Transformer MPVPE Test - #6 of 9 Archive leaderboard report
3D Interacting Hand Pose Estimation InterHand2.6M Keypoint Transformer MRRPE Test 29.63 #6 of 9 Archive leaderboard report
hand-object pose HO-3D v2 Keypoint-Trans ADD-S 21.4 #4 of 9 Archive leaderboard report
hand-object pose HO-3D v2 Keypoint-Trans Average MPJPE (mm) 25.5 #4 of 9 Archive leaderboard report
hand-object pose HO-3D v2 Keypoint-Trans OME 68.0 #4 of 9 Archive leaderboard report
hand-object pose HO-3D v2 Keypoint-Trans PA-MPJPE 10.8 #4 of 9 Archive leaderboard report
hand-object pose HO-3D v2 Keypoint-Trans ST-MPJPE 25.7 #4 of 9 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

Absolute Position EncodingsAdamAttentionBPEDense ConnectionsDropoutHeatmapLabel SmoothingLayer NormalizationLinear LayerMulti-Head AttentionPosition-Wise Feed-Forward LayerResidual ConnectionSoftmaxTransformer

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections