LOCATE: Localize and Transfer Object Parts for Weakly Supervised Affordance Grounding.
Gen LiVarun JampaniDeqing SunLaura Sevilla-LaraPublished in: CoRR (2023)
Keyphrases
- weakly supervised
- object class
- object categories
- object classes
- object detection
- object recognition
- viewpoint
- superpixels
- bounding box
- d objects
- object localization
- image features
- training examples
- computer vision
- image patches
- object detectors
- appearance model
- generative model
- training set
- partial occlusion
- natural images
- target object
- visual words
- object segmentation
- pairwise
- machine learning
- image content
- probabilistic model
- relation extraction
- image sequences