Seq2seq recurrent neural network applied to abstractive title generation of medical documents using both supervised learning and reinforcement learning formulations.
Robotic face tracking system that uses a deep learning model to detect faces and control a motorized webcam to keep detected faces centered in the field of view of the robot.
Convolutional Neural Network model predicting depth from single RGB images and deployed in real-time from a webcam feed. RGB-depth image pairs were collected automatically using a robot equipped with a 3D sensor.
