Computer Vision/Deployed
Real-time railway defect detection running YOLOv8 with a TensorRT INT8 engine, GStreamer hardware decode, and CUDA kernels on Jetson Orin.
TensorRT ONNX engine with INT8 quantization-aware training, GStreamer hardware-accelerated decode, and hand-written CUDA kernels. Histogram equalization handles the low-light and motion-blur conditions on real track footage.
INT8 QAT delivered 2× over the FP16 baseline while holding accuracy, quantization is an accuracy technique here, not a compromise.
Tell me what you're building, the constraints you're working with, and where it breaks. I reply within a day.