Computational commonalities of image understanding algorithms
Xavier Boix Bosch · Repository for Publications and Research Data (ETH Zurich) · 2015
A fundamental goal of Computer Vision is the automatic understanding of images.Semantic segmentation is a modality of image understanding, that aims at labeling each pixel of an image with a semantic class.In the literature, there is a plethora of semantic segmentation algorithms.Often these algorithms are engineered at the algorithmic and implementation levels, and they appear to be disconnected from prior art at the computational level.This has led to algorithms designed for specific applications and datasets.