Self-Attention blocks in UNet and FCN for accurate semantic segmentation of difficult object classes in autonomous driving.
Seyed-Hamid MousaviKin-Choong YowPublished in: CCECE (2023)
Keyphrases
- semantic segmentation
- object classes
- object categories
- object class
- pascal voc
- d objects
- image features
- superpixels
- conditional random fields
- scene classification
- object recognition
- training set
- image understanding
- object detection
- training examples
- high quality
- object segmentation
- bounding box
- computer vision
- weakly supervised
- generative model