HyperDense-Net: A hyper-densely connected CNN for multi-modal image semantic segmentation.
Jose DolzIsmail Ben AyedJing YuanChristian DesrosiersPublished in: CoRR (2017)
Keyphrases
- multi modal
- semantic segmentation
- input image
- image data
- superpixels
- scene classification
- image analysis
- image features
- image retrieval
- weakly supervised
- multiscale
- image representation
- image content
- image classification
- image collections
- pixel level
- conditional random fields
- image set
- region of interest
- similarity measure
- low level
- image segmentation
- image annotation
- image understanding
- image matching
- test images
- object class
- active contours
- high resolution
- high dimensional
- image processing