Improving 2D Human Pose Estimation in Rare Camera Views with Synthetic Data
This addresses a domain-specific limitation for computer vision applications that require pose estimation in non-standard camera angles, though it is incremental as it builds on existing synthetic data methods.
The paper tackled the problem of 2D human pose estimation in rare camera views (e.g., top- and bottom-view) by generating synthetic data with RePoGen, resulting in improved performance on these views without harming common-view accuracy, as shown in experiments on top-view datasets and a new real-image dataset.
Methods and datasets for human pose estimation focus predominantly on side- and front-view scenarios. We overcome the limitation by leveraging synthetic data and introduce RePoGen (RarE POses GENerator), an SMPL-based method for generating synthetic humans with comprehensive control over pose and view. Experiments on top-view datasets and a new dataset of real images with diverse poses show that adding the RePoGen data to the COCO dataset outperforms previous approaches to top- and bottom-view pose estimation without harming performance on common views. An ablation study shows that anatomical plausibility, a property prior research focused on, is not a prerequisite for effective performance. The introduced dataset and the corresponding code are available on https://mirapurkrabek.github.io/RePoGen-paper/ .