Completed

Cone library: Cone recognition

Published on the October 22, 2025 in IT & Programming

About this project

Open

The Cones Library is a computer vision and validation framework designed to analyze training drill videos by detecting and evaluating cone placement. It identifies all types of cones—standard or makeshift alternatives—regardless of shape, size, or color, and determines their exact locations, distances between them, and overall arrangement. The system verifies if cones are placed correctly according to official setup instructions, checking for straight lines or specific shapes (like squares or zig-zags), and uses linear regression and 3D modeling to assess accuracy. It also continuously monitors cones throughout the drill to ensure they remain unmoved, accounting for margin of error. If any cones are miscounted, misplaced, or moved, the video is flagged or disqualified. The framework is designed to be adaptable, with easily configurable thresholds to suit different drills and requirements.  The project would like to be completed as soon as possible.

Project overview

Direct Assist is a sports technology startup looking to take player performance and analysis to the next level by harnessing the power of AI.

Category IT & Programming
Subcategory Artificial Intelligence
Project size Small
Is this a project or a position? Project
Required availability As needed

Delivery term: Not specified

Skills needed