UperFormer: A Multi-scale Transformer-based Decoder for Semantic Segmentation.
Jing XuWentao ShiPan GaoZhengwei WangQizhu LiPublished in: CoRR (2022)
Keyphrases
- semantic segmentation
- multiscale
- superpixels
- street scenes
- conditional random fields
- weakly supervised
- label transfer
- natural images
- scene classification
- pixel level
- scale space
- image representation
- object class
- object classes
- image segmentation
- image understanding
- image processing
- object categories
- keypoints
- image fusion
- computer vision
- object recognition
- active learning
- pascal voc
- video sequences
- image classification
- motion estimation
- image features