Project Icon

sc_depth_pl

Comprehensive Self-supervised Monocular Depth Learning Approaches for Video Analysis

Product DescriptionSC-Depth offers a robust, three-version framework for self-supervised monocular depth estimation. V1 features geometry consistency and self-discovered masks for reliable depth predictions. V2 enhances indoor scene accuracy with an auto-rectify network managing rotation challenges. V3 uses pretrained networks for improved depth accuracy in dynamic scenes. With support for multiple datasets and comprehensive scripts, it suits diverse research and application needs.
Project Details