A Novel Mesh-Based 3D Mouse Pose Estimation with Low Jitter

Author(s)
Luo, Zhangqi
Advisor(s)
Editor(s)
Associated Organization(s)
Organizational Unit
Organizational Unit
School of Computer Science
School established in 2007
Supplementary to:
Abstract
Animal keypoint prediction is fundamental for neuroscience research. The existing off-shelf solution focus on the accuracy of keypoint prediction but doesn’t address the jitter. In this paper, we present a novel pose estimation method that leverages the 3D animal mesh and skeletons generated by SMAL model. We utilize animal 3D reconstruction results to stabilize the prediction of key points and reduces the jitter. Our approach is eventually experimented on mouse key points prediction.
Sponsor
Date
2024-08-20
Extent
Resource Type
Text
Resource Subtype
Thesis
Rights Statement
Rights URI