IMPose: Interactive Multi-person Pose Estimation with Dynamic Correction Propagation

2026-06-03

AI Summary

This work addresses the high cost of annotating multi-person dynamic poses and the inefficiency of existing tools in propagating cross-frame corrections in crowded scenes. The authors propose an interactive annotation method featuring a novel dual-level tracking mechanism that jointly operates at both keypoint and instance levels. By integrating a trajectory buffer to store historical information and employing a keypoint-aware embedding with relative positional encoding, the approach enables robust temporal propagation and cross-frame consistency. The method substantially improves annotation efficiency and quality under challenging conditions such as occlusion and motion blur: on 3DPW, it achieves high-precision labeling of 1,050 frames with only 27 user clicks; on PoseTrack21, it requires merely three clicks per 84-frame trajectory on average. Using just ten annotators over ten hours, the approach scales to produce 188K pose instances.

Abstract

High-quality dynamic human pose annotation equips AI with precise motion kinematics to enable human behavior mastery, yet remains labor-intensive and time-consuming. Current annotation tools either lack temporal correction propagation or fail in multi-person scenarios, necessitating excessive manual intervention. In this paper, we introduce IMPose, an interactive tool for multi-person dynamic pose annotation. It features a dual-level tracking mechanism that propagates one-frame multi-person pose corrections from annotators across entire videos. The keypoint-level ensures corrections temporal propagation via sequential modeling, while the instance-level employs keypoint-aware embedding with relative positional encoding to maintain multi-person cross-frame consistency. To further improve robustness, IMPose maintains historical pose and instance cues in a trajectory bank, which enhances long-range temporal association and stabilizes annotation in challenging cases such as occlusion and motion blur. By converting sparse human corrections into dense and coherent pose trajectories, our framework significantly reduces repeated manual refinement across frames. Extensive experiments show that IMPose consistently achieves a strong accuracy efficiency trade off under different interaction budgets, demonstrating particular advantages in low click annotation settings. IMPose achieves high precision annotation with high efficiency, requiring only 27 clicks per 1,050 frame video on 3DPW and 3 clicks per tracklet per 84-frame on PoseTrack21. We further expand PoseTrack21 with 188K pose instances (3.55M keypoints) at a minimal cost of 10 annotators in 10 hours. The annotation tool, codes, and extended dataset will be open-sourced.

Problem

• multi-person pose estimation
• temporal correction propagation
• dynamic pose annotation
• manual annotation efficiency
• occlusion and motion blur

Innovation

• interactive pose annotation
• multi-person pose estimation
• temporal correction propagation
• keypoint-aware embedding
• trajectory bank