Description
Build on the previous assignments where you captured a video-clip in real-time using a video camera. For this final project, please do the following:
- Work on real-time video (i.e., not captured video)
- Video should have few persons (say 2 to 4) moving around and at least one of them should be wearing UTD logo T-shirts.
- UTD T-shirts do have different types of UTD logos. You can choose to track any one (or more) of these logos. The choice of logos is yours. If UTD logo T-shirt is not available, you can use any T-shirt with a logo that is big and easy to recognize. But during the demo you should be wearing that logo T-shirt.
- Your task is to detect and track only the person wearing UTD T-shirt, i.e., put a bounding box on the entire person wearing the T-shirt (not just the T-shirt alone).
- Additionally, mark the tracked person’s face and eyes with different colored bounding boxes.
- A separate task for the project is to determine the height of the person in feet and inches. To do this you can use an object of known width and height to act as a calibration parameter. You need to track that object and the person, find the ratio and obtain the person’s real height. In order to do this, the person needs to be standing next to the object, at the same depth location.
- Person, face, eye, logo and object detection can be done using any OpenCV strategy – no restrictions.
Demo will be required to the TA. You should: (i) bring/wear the UTD logo T-shirt (or any other logo) that your program will track, (ii) bring the object with known dimensions for measuring the real world height.
Submission on eLearning (elearning.utdallas.edu)
Late Submissions
Late submissions NOT accepted.
Rubrics:
- Tracking (with bounding box) the person wearing UTD T-shirt (or any other logo) – 50%;
- Face and eye detection for only the tracked person (with bounding boxes) – 25%;
- Height of the person in real world dimensions – 25%