A Comparison of Directional Distances for Hand Pose Estimation
This work addresses the open problem of benchmarking for 3D hand tracking, which is incremental as it provides a new protocol and dataset for more accurate evaluation.
The authors tackled the problem of benchmarking 3D hand pose estimation by introducing a new dataset and protocol that avoids accumulative error, enabling separate error measurement for frame pairs of varying difficulty. They evaluated directional distances for silhouette-based tracking, identifying the best-performing method through a comparative study.
Benchmarking methods for 3d hand tracking is still an open problem due to the difficulty of acquiring ground truth data. We introduce a new dataset and benchmarking protocol that is insensitive to the accumulative error of other protocols. To this end, we create testing frame pairs of increasing difficulty and measure the pose estimation error separately for each of them. This approach gives new insights and allows to accurately study the performance of each feature or method without employing a full tracking pipeline. Following this protocol, we evaluate various directional distances in the context of silhouette-based 3d hand tracking, expressed as special cases of a generalized Chamfer distance form. An appropriate parameter setup is proposed for each of them, and a comparative study reveals the best performing method in this context.