| Group |
2021-15 |
Status |
completed |
| Title |
Person Detection from Crowded Video Images |
| Supervisor |
Maria Amer |
| Description |
Visual detection and identification of an individual in a crowded environment observed by a camera is important in variety of applications including space management, border control, and surveillance. Person detection from video images is challenging due to spatial and temporal visual feature variations and due to strong visual similarity in persons’ appearance. Often the images are of low-resolution and poor quality.
In this project, the students will design and implement a system for person detection from video data of crowded scenes. The students will then integrate this system into a computer vision application such as human pose estimation (https://github.com/facebookresearch/detectron2). Human pose estimation is important for higher level reasoning in human-computer interaction and activity recognition (such as the recognition of falling person). Students may select another application of their choice.
This project is a complex multi-disciplinary design problem involving communications, signal processing, computer vision, artificial intelligence, and software design.
|
| Student Requirement |
Signals and Systems, Signal processing, Image processing, Programming (C/C++, Python, Matlab) |
| Tools |
Linux, OpenCV, Deep learning packages (such as Tensorflow and Pytorch). |
| Number of Students |
6 |
| Students |
Maria Kovalenko, Mayah Leonce, Arseny Kokotov, Amber Augustus,
Anthony Biancardi-Maffei, Fatmah Almaas |
| Comments: |
|
| Links: |
|