Occlusion-Aware Multi-Target Detection and Tracking

    Project: Research

    Project Details

    Description

    This project aims to explore multi-scale single-shot detectors for multi-subject face detection and tracking, with a particular focus on the "multi-subject and robustness to facial occlusion. We will mainly develop deep learning technologies, such as the convolutional neural network (CNN), to obtain more robust features. Furthermore, in order to handle occlusions, we will study data augmentation techniques so as to enrich the training set. These techniques potentially include the generative adversarial network and high-fidelity data synthesis. The project will last about a period of 6 months and will be directed by Dr Liang Zheng, Computer Science Futures Fellow, and A/Prof Stephen Gould, College of Engineering and Computer Science (CECS), Australian National University (ANU). The collaboration partner is Dr. Akshay Asthana, Senior Staff Algorithm Scientist with the Seeing Machines.
    StatusFinished
    Effective start/end date19/12/1930/06/20

    Fingerprint

    Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.