COIN: Counterfactual inpainting for weakly supervised semantic segmentation for medical images.
Dmytro ShvetsovJoonas ArivaMarharyta DomnichRaul VicenteDmytro FishmanPublished in: CoRR (2024)
Keyphrases
- medical images
- weakly supervised
- semantic segmentation
- superpixels
- object class
- deformable models
- topic models
- relation extraction
- named entities
- semi supervised
- region of interest
- co occurrence
- denoising
- viewpoint
- object classes
- computer vision
- object detection
- natural images
- multiple objects
- bounding box
- information extraction
- image data
- object detectors
- probabilistic model
- active learning
- higher order