Benchmarking Segmentation Models with Mask-Preserved Attribute Editing
Zijin Yin, Kongming Liang, Bing Li, Zhanyu Ma, Jun Guo
CVPR 2024
This repository contains the code for "Benchmarking Segmentation Models with Mask-Preserved Attribute Editing", a framework editing visual attributes for robustness evaluation of segmentation models.
- Environment setup
- Code for generating attribute variations
- Our Pascal-EA(Editable Attributes) benchmark