Five-model DeepStream pipeline on a Jetson Orin Nano that boots to a live dashboard in 15 seconds
Five sequential AI models — PeopleNet, NvDCF tracker, MoveNet pose, YOLOX hand detection, MediaPipe hand landmarks — running natively on a Jetson Orin Nano. The trick to skipping the 12-minute first-boot TensorRT compile: ship the engines inside the OTA payload, not just the ONNX.