SemFormer: Semantic Guided Activation Transformer for Weakly Supervised Semantic Segmentation.
Junliang ChenXiaodong ZhaoCheng LuoLinlin ShenPublished in: CoRR (2022)
Keyphrases
- weakly supervised
- semantic segmentation
- superpixels
- relation extraction
- object class
- topic models
- semantic features
- domain specific
- named entities
- conditional random fields
- scene classification
- multiscale
- semi supervised
- semantic information
- object recognition
- image representation
- low level features
- higher order
- long range
- visual features
- automatic extraction
- viewpoint
- natural language