Residual Multiscale Full Convolutional Network (RM-FCN) for High Resolution Semantic Segmentation of Retinal Vasculature.
Tariq Mahmood KhanAntonio Robles-KellySyed Saud NaqviMuhammad ArsalanPublished in: S+SSPR (2020)
Keyphrases
- semantic segmentation
- multiscale
- convolutional network
- superpixels
- high resolution
- coarse to fine
- convolutional neural networks
- image processing
- low resolution
- weakly supervised
- super resolution
- conditional random fields
- natural images
- scale space
- object class
- image representation
- scene classification
- remote sensing
- image segmentation
- object categories
- object detection
- edge detection
- segmentation algorithm
- image analysis
- object segmentation
- multiresolution
- keypoints
- depth map
- face images
- image patches
- medical images
- input image