Bending Reality: Distortion-aware Transformers for Adapting to Panoramic Semantic Segmentation.
Jiaming ZhangKailun YangChaoxiang MaSimon ReißKunyu PengRainer StiefelhagenPublished in: CVPR (2022)
Keyphrases
- semantic segmentation
- street scenes
- conditional random fields
- superpixels
- label transfer
- weakly supervised
- scene classification
- object class
- object categories
- object classes
- pascal voc
- multiscale
- object detection
- image set
- image understanding
- viewpoint
- higher order
- input image
- feature extraction
- small number
- n gram
- segmentation algorithm
- generative model
- natural images
- pairwise