Yuxing Yang
Discovering anomalies in video sequences.
Email: y.yang60@ncl.ac.uk
Project title
Video-based human tracking and action recognition
Supervisors
Project description
We are detecting whether there are anomaly varieties or actions in video sequences. We are using methods for recognition of human targets.
Methodology and objectives
The basic neural networks in computer vision are:
- the convolutional neural network
- the recurrent neural network
The main tasks of action classification are:
- multiple human tracking
- human action recognition
Both use a deep neural network. Our research will detect video anomaly detection. We have based the method on a deep convolutional generative adversarial network.
Result
The results can be used in video object or action classification to discover anomalies in video sequences. They can also perform intelligent monitoring in real time.
Interests
Computer Vision. Action Recognition. Anomaly Detection.
Qualifications
MSc in Communication and Signal Processing from Newcastle University