AI Object Detection System Build

I’m diving deep into AI and Machine Learning with a clear objective: design and implement a robust Computer Vision pipeline dedicated to accurate, real-time Object Detection. I already have several experimental datasets and a few early ideas, but I need an experienced partner who can turn these concepts into a production-ready solution. What I’m after • Model research & selection: advise on the most suitable architectures (YOLO-v8, Faster R-CNN, DETR, or anything you feel is a smarter fit). • End-to-end development: data preprocessing, augmentation, training, validation, and hyper-parameter tuning using Python with TensorFlow or PyTorch plus essential OpenCV utilities. • Performance optimisation: achieve fast inference on both GPU and edge devices (CUDA / TensorRT knowledge is a big plus). • Deployment wrapper: export to ONNX or TensorFlow-Lite and integrate a simple API or demo app so I can test detections live. • Knowledge transfer: clean, well-commented code in a private repo, plus a short hand-off session (recorded or live) so future iterations are straightforward. Acceptance criteria – mAP and FPS targets agreed upfront and met on a held-out test set. – Reproducible training notebook or script that runs on a standard GPU instance. – Lightweight demo (CLI or minimal GUI) proving real-time detection on sample video. – All original source files, environment specs, and a brief implementation report delivered. If you’re ready to push the boundaries of Computer Vision and enjoy the thrill of seeing objects pop into bounding boxes with millisecond accuracy, let’s build this together.