Discovering Useful Parts for Pose Estimation in Sparsely Annotated Datasets
This work addresses pose estimation for applications like biological motion analysis, but it is incremental as it builds on existing part-based models.
The paper tackles the problem of improving pose estimation accuracy by discovering parts from unannotated image regions, which are used to enhance appearance likelihoods in part-based models. The result shows that the approach localizes landmarks at least twice as accurately as a baseline Mixture of Pictorial Structures model on a hawkmoth flight dataset.
Our work introduces a novel way to increase pose estimation accuracy by discovering parts from unannotated regions of training images. Discovered parts are used to generate more accurate appearance likelihoods for traditional part-based models like Pictorial Structures [13] and its derivatives. Our experiments on images of a hawkmoth in flight show that our proposed approach significantly improves over existing work [27] for this application, while also being more generally applicable. Our proposed approach localizes landmarks at least twice as accurately as a baseline based on a Mixture of Pictorial Structures (MPS) model. Our unique High-Resolution Moth Flight (HRMF) dataset is made publicly available with annotations.