[深度学习论文笔记][Semantic Segmentation] Learning Hierarchical Features for Scene Labeling

Farabet, Clement, et al. “Learning hierarchical features for scene labeling.” IEEE transactions on pattern analysis and machine intelligence 35.8 (2013): 1915-1929. (Citations:

703).

1 Pipeline
See Fig. We use multiple scales of the image to capture both short-range information and long-range information, which turns out to beat the single scale version by a large

margin. The network of different scales are the same. Besides, traditional methods like superpixels or segmentation tree are computed to exploit the natural contours of the image.