# Hello AI World — Jetson Nano Deep Learning Hello AI World is NVIDIA's official introductory tutorial for on-device AI inference on Jetson hardware. It provides a collection of DNN-based vision applications — image classification, object detection, semantic segmentation, depth estimation — that run in real time on the Jetson's GPU using the TensorRT inference framework. Every model ships as a pre-built TensorRT engine or can be re-trained on custom data via PyTorch and then re-exported. The entire workflow (data → train → export → infer) runs on the Jetson itself. ## Applications Included | App | Task | Backbone | |-----|------|----------| | `imagenet` | Image classification | ResNet-18 / GoogLeNet | | `detectnet` | Object detection | SSD-Mobilenet-v2 / PedNet | | `segnet` | Semantic segmentation | FCN-ResNet18 | | `depthnet` | Monocular depth | MonoDepth2 | | `posenet` | Human pose estimation | ResNet-18 | ## Hardware - NVIDIA Jetson Nano (B01) — 4GB preferred - USB or CSI camera - 5V/4A power supply (barrel jack) ## Source - Repo: https://github.com/dusty-nv/jetson-inference - License: MIT - Docs: https://github.com/dusty-nv/jetson-inference/blob/master/docs/
Category: Other Robots
| Item | Qty | Notes | |------|-----|-------| | NVIDIA Jetson Nano (B01, 4GB) | 1 | 4GB preferred for larger TRT models | | USB webcam or CSI camera | 1 | Inference input | | 5V/4A power supply (barrel jack) | 1 | Required — USB OTG power insufficient | | microSD card (32 GB+, A2 class) | 1 | For JetPack + models |