VEIL: Vetting Extracted Image Labels from In-the-Wild Captions for Weakly-Supervised Object Detection.
Arushi RaiAdriana KovashkaPublished in: CoRR (2023)
Keyphrases
- weakly supervised
- weakly labeled
- semantic segmentation
- object detection
- object detectors
- object class
- image data
- image features
- image content
- fully supervised
- multiscale
- region of interest
- input image
- single image
- image segmentation
- image retrieval
- object categories
- image representation
- image classification
- relation extraction
- low level
- superpixels
- bounding box
- topic models
- lighting conditions
- image regions
- segmentation method
- multi class
- viewpoint
- pairwise
- object recognition
- training data