Скачать с ютуб quarter CNN: Faster R-CNN в хорошем качестве

quarter CNN: Faster R-CNN

This lecture introduces an object detector called Faster R-CNN. It was proposed in the paper entitled as "Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Network”, NIPS, 2015, and IEEE Trans. on PAMI, 2017. The outline of this lecture includes: (1) Introduction about what object detection is, the Faster R-CNN background, and its architecture overview. (2) Network ingredients about the components of Faster R-CNN including region proposal network, RoI pooling, and detection network. (3)Faster R-CNN training about the training steps and two defined loss functions including RPN loss and detection loss. 0:00 Outline 1:12 About Object Detection 1:49 Background 3:25 Architecture Overview 6:58 Region Proposal Network(RPN) 11:01 RoI Pooling 12:15 Detection Network 14:57 Training Overview 16:41 RPN Loss 20:13 Detection Loss Any comments are welcome. (email: [email protected]) All resources are available on the website (http://quarter.tw)