High-fidelity Pseudo-labels for Boosting Weakly-Supervised Segmentation.
Arvi JonnarthYushan ZhangMichael FelsbergPublished in: CoRR (2023)
Keyphrases
- weakly supervised
- high fidelity
- semantic segmentation
- fully supervised
- superpixels
- object class
- real time
- object detectors
- weakly labeled
- segmentation algorithm
- relation extraction
- object segmentation
- semi supervised
- topic models
- conditional random fields
- named entities
- high quality
- segmentation method
- object recognition
- multiscale
- image segmentation
- labeled data
- level set
- pairwise
- training data
- object categories
- graph structure
- bounding box
- learning algorithm