Multi-Object Tracking in basketball videos using Deep Learning.
Goal: Applying GRU (= Gated Recurrent Unit) in order to track basketball players in videos.
Programming language: Python
Steps that I think that need to be taken:
1. Extract frames from video at timestamps specified the labels
2. Detect objects that need to be tracked using a pre-trained network (e.g., ResNet-50, VGG16 or alike)
3. Match detections with ground_truth bounding box coordinates
4. Apply GRU to track the players in multiple frames and predict the position of the players in frame t + 1.
5. Evaluate the performance using a MOT-2D benchmark
video to be tracked: [login to view URL]
Attachment: bounding box coordinates of each player at a certain time.
Contains the following columns:
- column 1: YouTube ID
- column 2: Time corresponding to the video frame, in microsec.
- column 3: Top-left x-coordinate of player bounding box in frame relative to frame width.
- column 4: Top-left y-coordinate of player bounding box in frame relative to height
- column 5: Width of player bounding box relative to frame width.
- column 6: Height of player bounding box relative to frame height.
- column 7: A player ID
11 freelancere byder i gennemsnit €271 på dette job
I DO NOT OUTSOURCE I have been a freelancer for the past 8 years, I believe that my experience and skill in this background will prove to be of great help to you. Contact me to discuss more on the details