PEARL: Preprocessing Enhanced Adversarial Robust Learning of Image Deraining for Semantic Segmentation.
Xianghao JiaoYaohua LiuJiaxin GaoXinyuan ChuRisheng LiuXin FanPublished in: CoRR (2023)
Keyphrases
- semantic segmentation
- weakly supervised
- input image
- single image
- learning algorithm
- image content
- image data
- supervised learning
- multiscale
- superpixels
- image classification
- image features
- high resolution
- conditional random fields
- image matching
- scene classification
- image retrieval
- active learning
- test images
- object categories
- partial occlusion
- bounding box
- pixel level
- viewpoint