Project Info

ScoutIT

Devpost

Inspiration

"What started with a winning strategy —finding undervalued players while others chased the obvious choices—evolved into a bigger revelation. Like Moneyball revolutionized baseball through data, we saw an opportunity to transform scouting itself. But we're going beyond just statistics. Through computer vision and AI, we're capturing the intangibles: the perfect arc of a shot, the millisecond decisions, the biomechanics that make athletes exceptional. This isn't just an iteration of scouting—it's a complete reimagining." What It Does ScoutIT combines cutting-edge AI with sports science to deliver two game-changing features: Quick Evaluation: Upload any player's footage, and our AI instantly compares their mechanics against a database of elite athletes, providing immediate insights and improvement areas. Potential Predictor: Using our proprietary Differences in Differences Model powered by GenAI, we analyze a player's progression over time to forecast their peak potential, considering factors traditional scouts might miss. How We Built It Our development approach combined the following key technologies Real-time Tracking and Masking: Implementing YOLO object detection models to follow ball movement and player interactions based on which we perform our image masking. Movement Analysis through Pose Estimation: Leveraging OpenPose and MediaPipe to extract precise skeletal data enabling us to analyze deeper biomechanics Vector Image Reconstruction: Developing diffusion models into our processing pipeline reconstructing uploading videos to create an "ideal shot video" path based on which our LLM models are finetuned Performance Scoring: Creating a standardized 100-point scoring system based on Hall of Famers like Stephen Curry and LeBron James Growth Prediction through Vision Models: Using the finetuned GenAI vision models to generate feature importance arrays that project a player's development trajectory Challenges We Ran Into Our biggest hurdles became our greatest innovations: Data Variability: Developed adaptive preprocessing algorithms to handle diverse video qualities and conditions Player Tracking and Masking: Created a ball-focused tracking system that maintains player identification in crowded situations by incorporating Convex Hull masking techniques Insight Generation: Implemented GenAI to translate complex metrics into actionable basketball insights" What's Next for ScoutIT? ScoutIT is set to transform sports scouting across three dimensions: Multi-Sport Evolution: Adapting our AI to analyze sport-specific mechanics across basketball, soccer, baseball, and esports Real-Time Intelligence: Developing instant analysis capabilities for in-game strategic adjustments Democratizing Development: Creating personalized training insights accessible to athletes at all levels" We believe the sky is the limit for ScoutIT! Impact ScoutIT isn't just changing how we scout talent—it's democratizing access to professional-level analysis. By making advanced biomechanical analysis accessible to teams and athletes at all levels, we're helping uncover hidden talent and potential that traditional scouting might miss. This technology could be particularly transformative for under-resourced programs and emerging sports markets.

Analysis

Compare with all teams

View

Metric

Figures cover GitHub contributors during the hackathon window. A co-authored commit counts in full for each author, so per-member totals add up to more than the whole-team figures.

Technology

Found in codeClaimed only
  • CIn code
  • C++In code
  • CSSIn code
  • FastAPIIn code
  • FirebaseIn code
  • HTMLIn code
  • JavaScriptIn code
  • LangChainIn code
  • PythonIn code
  • PyTorchIn code
  • ReactIn code
  • Tailwind CSSIn code
  • TypeScriptIn code
  • Google GeminiClaimed
  • StreamlitClaimed

13 of 15 appear in the indexed code. 2 claimed on Devpost could not be matched to code, which may simply mean the tool leaves no trace in the repository.

AI coding agents

No AI coding agent signals were found in this repository.

Detected from committed agent config files and commit authorship. Absence of a signal is not proof an agent was unused.

Codebase size

Source size

26 MB

Source files

1,262

Counts recognized source files only; vendored directories, binaries and lockfiles are excluded, so this is smaller than the repository on disk.

0 stars