Convolutional neural networks have been the dominant force in computer vision for a very long time. In this talk, we start from the very basics of convolutional networks and then discuss how CNNs have evolved from classifying individual objects to detecting multiple objects on a picture. We will consider in detail several well-known object detection models, including YoLo, single-shot detectors, and a range of models from R-CNN to the recently released Mask R-CNN for segmentation.
Download presentation