Comparison of Spatio-Temporal Hand Pose Denoising Models
Published in Universitat de Barcelona, 2022
Human pose estimation in video sequences presents unique challenges due to temporal context — noise in keypoint detection, incomplete skeletons, and temporal inconsistencies across frames. This thesis systematically compares spatio-temporal models for hand pose denoising, evaluating their effectiveness in refining noisy pose estimates from video data.
Key contributions:
- Comprehensive comparison of denoising architectures for hand pose refinement
- Analysis of temporal coherence methods for video-based pose estimation
- Evaluation framework applicable to broader human pose estimation tasks
This work laid the foundation for my PhD research in Human-Robot-Object Interactions, where accurate hand and body pose understanding is critical for enabling robots to interpret and respond to human actions.
Recommended citation: Johnny Núñez. (2022). "Comparison of Spatio-Temporal Hand Pose Denoising Models." Universitat de Barcelona.
Download Paper | Download Slides