DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs.

Autor: Chen, Liang-Chieh, Papandreou, George, Kokkinos, Iasonas, Murphy, Kevin, Yuille, Alan L.
Předmět:
Zdroj: IEEE Transactions on Pattern Analysis & Machine Intelligence; Apr2018, Vol. 40 Issue 4, p834-848, 15p
Abstrakt: The article presents information on semantic image segmentation with deep learning. According to author, atrous convolution enables one to directly control the resolution at which feature responses are calculated in case of deep convolutional neural networks. One-dimensional as well as two-dimensional illustrations of arouse convolutions are also presented.
Databáze: Complementary Index