Segment scene with semi-supervised learning using monocular depth estimation with DG
Segment scene with semi-supervised learning using monocular depth estimation with DG
Depth-Guided Semi-Supervised Instance Segmentation
arXiv paper abstract https://arxiv.org/abs/2406.17413
arXiv PDF paper https://arxiv.org/pdf/2406.17413
Semi-Supervised Instance Segmentation (SSIS) aims to leverage an amount of unlabeled data during training.
… introduce a Depth-Guided (DG) SSIS framework … uses depth maps … from input images, which represent … instances with … distance values, offering … contours for … instances.
… propose Depth Feature Fusion, which integrates features … from depth estimation … allows the model to understand depth … better and ensure its effective utilization.
… to manage the variability of depth images during training, … introduce the Depth Controller.
This component enables adaptive adjustments of the depth map, enhancing convergence speed and dynamically balancing the loss weights between RGB and depth maps.
… approach establishes a new benchmark for SSIS, outperforming previous methods …
Stay up to date. Subscribe to my posts https://morrislee1234.wixsite.com/website/contact
Web site with my other posts by category https://morrislee1234.wixsite.com/website
LinkedIn https://www.linkedin.com/in/morris-lee-47877b7b
