GSVNet: Guided Spatially-Varying Convolution for Fast Semantic Segmentation on Video.

Shih-Po Lee Si-Cun Chen Wen-Hsiao Peng

Published in: CoRR (2021)

Keyphrases

spatially varying
semantic segmentation
photometric stereo
superpixels
video data
video sequences
conditional random fields
noisy images
blurred images
motion blur
object categories
video frames
high quality
lighting conditions
image processing
object classes
space time
viewpoint