I will build, integrate ocr app or ai object detection mobile app, software, or web app


About this gig
I'll deliver a fast, reliable Flutter mobile app or backend software with integrated AI sports object detection (players, ball, referees, poses) and OCR for scoreboards, jerseys, or overlays. I build production-ready solutions
Services offered:
- Real-time sports object detection (YOLO / EfficientDet / custom models)
- Pose estimation & player tracking (OpenPose / MediaPipe)
- OCR integration for scoreboards, timestamps, jersey numbers (Tesseract / ML Kit / Vision)
- Flutter app integration (iOS & Android)
- On-device inference (TFLite / CoreML) or server inference (PyTorch/ONNX)
- Video stream processing & frame sampling
- Multi-camera / live-stream ingestion support
- Event detection (goal, foul, out-of-bounds) & analytics hooks
- Data export (CSV/JSON), WebSocket or REST API endpoints
- Heatmaps, tracking overlays, and visual debug UI
- Model training / fine-tuning on custom sports datasets
- Performance tuning: quantization, pruning, batching
- CI/CD, testing, and app store deployment support
Why choose this service?
- Real-time, mobile-optimized AI pipelines
- Full Flutter integration
- Custom model training & high accuracy focus
- Clear delivery code, models, docs, and deployment
Contact me and let's start
Get to know Ibro T
Full Stack Development, OpenCV, WebApp, AR, VR App, and AI Development
- FromUnited States
- Member sinceOct 2025
Languages
English, Spanish, Arabic, German
FAQ
Tech & Tools I Use
YOLOv5 / YOLOv8 / EfficientDet / Detectron2 TensorFlow Lite / Core ML / ONNX Runtime MediaPipe / OpenPose for pose estimation Tesseract OCR / Google ML Kit / Apple Vision Flutter (Dart), Platform channels (Swift/Kotlin) Python, PyTorch, TensorFlow, FastAPI / Flask Firebase / AWS / GCP integratings
Do you need my video/data to start?
For best accuracy yes, provide sample footage, labels, or I can use public datasets and offer custom dataset labeling for a fee.
Will detection run offline on the phone?
Yes, on-device TFLite/CoreML inference is available; server inference is optional for heavier models.
Can you detect specific sports/events?
Yes, I can fine-tune models for football, basketball, tennis, baseball, and custom event detection (goal, foul, serve, etc.).
How do you handle latency and performance?
I optimize via model pruning, quantization, frame sampling, and efficient pipelines; I’ll test across real devices.

