Portfolio

Display detailed information about your work

ShotPage

ShotPage is one of our own products that's currently in development. Tee'd up on ShotPage.com, ShotPage is an App and Web based sports video analysis business for tennis players and golfers (more sports to be added at later times). ShotPage allows a user to process raw video footage of them performing their shots and then creates 'Shot Clips' out of the raw video. These shot clips are all 3 seconds in length (1.5 secs before the contact point and 1.5 secs after) meaning every clip is standardised with the contact point in the middle. As well as the 3 second video clip are 21 still images generated of the shot (10 before contact, the contact frame, 10 after contact). These 21 still images are time coded to be captured at the most important points of the shot. The video and 21 still images are bundled together to form the Shot Clip.

The project will be launched in Phase 1 in the above format and clips can be analysed by coaches to allow players to get tips on their game. As we grow the dataset of captured shots, we will process this dataset into a machine learning model for "AI Coaching". The dataset will be curated by professional coaches and annotated with tips of the given shot in the dataset. The AI model will then be trained and be able to determine coaching techniques for improving shot technique as a result of the generated AI model, effectively allowing SHotPage to be your "AI Coach" helping you refine your shot technique.

Client
Exon AI Ltd
AI technology
Computer Vision: Object Detection, Semantic Segmentation
Date
08 Jan 2017
Download
Share