I’m building a mobile overlay for 8-ball that draws extended aiming lines over the table and instantly highlights every ball-to-cushion or ball-to-ball collision. The end result should run natively on both Android and iOS; I’m not interested in a browser plug-in or desktop extension. Core requirements • Real-time collision detection powered by OpenCV (or an equally capable computer-vision library). • Cross-platform build—one shared codebase that compiles cleanly for Android and iOS. • Custom ML model trained to understand ball layout and table boundaries. Dataset & training I do not yet have training data. You’ll need to help me capture and label: • High-resolution images of ball and table positions. • Recorded game videos covering a variety of lighting angles and table colors. The HTML5 demo below shows the game logic I want to mirror visually: https://preview.codecanyon.net/item/8-ball-pro-html5-sport-game/full_screen_preview/37375896 Deliverables 1. A working prototype that overlays long aiming lines and collision markers in real time. 2. The curated dataset and clear documentation of labeling conventions. 3. Training scripts plus the final model weights. 4. Xcode and Android Studio projects, fully commented, ready for store submission. Nice-to-have (not required but mention if you can cover them): 3D visualisation or game-state saving. If you’re experienced with OpenCV, mobile AR, and ML data pipelines, I’d love to hear how you’d tackle capture, labeling, and frame-rate optimization on mid-range devices.