Towards closing the gap in weakly supervised semantic segmentation with DCNNs: Combining local and global models.
Christoph MayerRadu TimofteGrégory PaulPublished in: Comput. Vis. Image Underst. (2021)
Keyphrases
- weakly supervised
- semantic segmentation
- superpixels
- topic models
- relation extraction
- conditional random fields
- semi supervised
- probabilistic model
- pascal voc
- scene classification
- object class
- named entities
- co occurrence
- input image
- sparse representation
- natural images
- multi class
- information extraction
- learning algorithm