Improve training of object segmentation with only boxes by using mask quality with BoxTeacher
Improve training of object segmentation with only boxes by using mask quality with BoxTeacher
BoxTeacher: Exploring High-Quality Pseudo Labels for Weakly Supervised Instance Segmentation
arXiv paper abstract https://arxiv.org/abs/2210.05174v1
arXiv PDF paper https://arxiv.org/pdf/2210.05174v1.pdf
Labeling objects with pixel-wise segmentation requires a huge amount of human labor compared to bounding boxes.
Most existing methods for weakly supervised instance segmentation focus on designing heuristic losses with priors from bounding boxes.
While … find that box-supervised methods can produce some fine segmentation masks and … wonder whether the detectors could learn from these fine masks while ignoring low-quality masks.
… present BoxTeacher, an efficient and end-to-end training framework … which leverages a sophisticated teacher to generate high-quality masks as pseudo labels.
… estimate the quality of pseudo masks, and propose the noise-aware pixel loss and noise-reduced affinity loss to adaptively optimize the student with pseudo masks.
… achieves 34.4 mask AP and 35.4 mask AP with ResNet-50 and ResNet-101 … which outperforms the previous state-of-the-art methods by a significant margin …
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
