adithyanraj03/CourtKeyNet
CourtKeyNet is an open source deep learning model for high fidelity badminton court detection, featuring an Octave Feature Extractor, Polar Transform Attention, and Geometric Consistency analysis
What's novel
CourtKeyNet is an open source deep learning model for high fidelity badminton court detection, featuring an Octave Feature Extractor, Polar Transform Attention, and Geometric Consistency analysis
Code Analysis
12 files read · 4 roundsA deep learning model for badminton court corner keypoint detection using a novel octave-based multi-scale feature extractor, polar transform attention, and geometric constraint module to ensure valid quadrilateral outputs.
Strengths
Genuinely novel architecture with multiple innovative components (octave feature extraction, polar attention, geometric constraints) and a complete end-to-end system including training, inference with confidence scoring, and evaluation metrics.
Weaknesses
No test suite, geometric losses disabled by default in main training config, hardcoded paths and confidence thresholds, and the inference GUI is large and could benefit from refactoring.
Score Breakdown
Signal breakdown
Innovation
Craft
Traction
Scope
Evidence
Commits
28
Contributors
1
Files
70
Active weeks
5
Repository
Language
Python
Stars
3
Forks
0
License
MIT