In computer vision, there are three main tasks, classification, detection and segmentation.
General classification as a backbone and there, in the last article, we introduced the model over 14 years, the details may refer to: target detection over the years the best model
This article deals introduction segmentation model.
In the course of the study semantic segmentation, the depth of learning has been occupy a major position. , The current two main open-source data sets PASCALVOC and IMAGENET by building different models of network test and evaluate, already has become a new direction.
Next, the current mainstream network segmentation as a summary to see when needed:
RefineNet: [CVPR. 17]: the Multi-Refinement Networks, the Path for High-Resolution the Semantic Segmentation
details, refer to: semantic segmentation of RefineNet
To be updated, wait a moment. . .