Video Data Pose Collection Pipeline

Video Preprocessing Pipeline using YoloV8 to crop/segment people out of videos and performing Mediapipe pose detection on them.
Current Issues:

- [ ] Configure jupyter notebook to utilize my personal gpu with the right Nvidia drivers
- [ ] Mediapipe Pose function: If part of a person is occluded, replace those coordinates with zeros and make sure the final shape of our Mediapipe function is consistent
- [ ] Debugging Proper cropping
- [ ] Function to validate correct person's coordinates are being extracted, based on our dataset and their coordinates from it. Also deal with edge case where if when cropping a background person's coordinates aren't being extracted as opposed to the foreground, main person.
- [ ] Experiment with segmentation to see if it creates too much of a performance hit versus accuracy of pose detection