Human poses admit complicated articulations and multi-granular similarity. Previous work on learning human pose metric utilize sparse models, which concentrate large weights on highly close poses and fail to depict an overall structure of human poses with multi-granular similarity. Moreover, previous work require a large number of similar/dissimilar annotated pairwise poses, which is an tedious task and remains inaccurate due to different subjective judgements of experts. Motivated by graph-based neighbor assignment techniques, we propose an unsupervised model called Sparsity Locality Preserving Projection with Adaptive Neighbors (SLPPAN), for learning human pose distance metric. By using a locality preserving penalty and a property of the graph Laplacian, SLPPAN introduces a fixed-rank constraint to enforce correct graph structure of poses, and learns the neighbor assignment, the similarity measurement and pose metric simultaneously. Different from previous work which require a great amount of similar/dissimilar annotated pairwise poses, the locality preserving penalty explores the underlying structure of poses in an unsupervised fashion. Experiments on pose retrieval of CMU Mocap database demonstrate that SLPPAN outperforms traditional pose metric learning methods by capturing view-point variations of human poses. Experiments on keyframe extraction of MSRAction3D database demonstrate that SLPPAN outperforms current methods by precisely detecting important frames of action sequences, even with a large amount of noise.
To View the Abstract Contents
Or Enquire Now !!!, WISEN Project Specialist will contact you soon.
Now it is Your Time to Shine.
Great careers Start Here.
We Guide you to Every Step
Success! You're Awesome
Thank you for filling out your information!
We’ve sent you an email with your Final Year Project PPT file download link at the email address you provided. Please enjoy, and let us know if there’s anything else we can help you with.
To know more details Call 900 31 31 555
The WISEN Team