In this technical talk, Eyal will discuss the application of deep learning to semantic segmentation. He will discuss datasets, evaluation metrics, and losses, and review the architectures and training methods of the main papers of 2014-2017, including: FCN, DeepLab, DeconvNet, U-Net, SegNet, Dilated Convolutions, 100-Layer Tiramisu, Wide ResNet, PSPNet, Adversarial methods, PolygonRNN, Mask R-CNN and semi-supervised methods. Notice that this talks assumes familiarity with convnets.
Hall: Hall I
Track: General Deep Learning