CLOT: Closed Loop Optimal Transport for Unsupervised Action Segmentation

Key Points

• The aim is to develop a closed loop optimal transport method for unsupervised action segmentation in video data.
• Utilizes closed loop optimal transport for data alignment and segmentation.
• Implements unsupervised learning techniques to analyze video action sequences.
• Evaluates performance against existing unsupervised action segmentation models.
• Achieves improved segmentation accuracy compared to baseline models.
• Demonstrates efficiency in processing complex video data sequences.
• Validates the effectiveness of the closed loop approach in real-world applications.