Ghost-UNet: An Asymmetric Encoder-Decoder Architecture for Semantic Segmentation From Scratch.
Iman Abaspur KazerouniGerard DoolyDaniel ToalPublished in: IEEE Access (2021)
Keyphrases
- semantic segmentation
- street scenes
- video codec
- distributed video coding
- conditional random fields
- superpixels
- label transfer
- rate distortion
- scene classification
- weakly supervised
- motion estimation
- pascal voc
- object detection
- object categories
- object classes
- object class
- computer vision
- video sequences
- viewpoint
- bit rate
- video coding
- pixel level
- wyner ziv
- image understanding