The Jetson platform is an extremely powerful way to begin learning about or implementing deep learning computing into your project. Start building a deep learning neural network quickly with NVIDIA's Jetson TX1 or TX2 Development Kits or Modules and this Deep Vision Tutorial.
What is Deep Learning?
The two main phases of deep learning network development are training and inference. NVIDIA provides a concise introduction to both in their Explanation of Deep Learning tutorial.
-
Deep Learning System Setup
-
NVIDIA Image Recognition
-
NVIDIA Digits Object Detection
-
Segmentation
This tutorial uses a host PC for training the Deep Neural Networks (DNNs) alongside a Jetson for inference. The host PC is used to flash the Jetson device with the current JetPack SDK version. Also learn how to install Ubuntu on the Host, how to run JetPack on the Host, install NVIDIA drivers and more!
Use NVIDIA’s ‘imageNet’ to begin recognizing specific images and representing them as a probability of being a desired object. NVIDIA provides a live camera program, a command line interface and a database of 1000 pre-trained objects to use with AlexNet and GoogleNet networks.
Learn how to use NVIDIA's ‘detectNet’ object detection / localization network to physically locate objects being seen by the camera rather than just identifying their probability of being a specific object.
Learn how to use a pre-trained ‘segNet’ recognition model for environmental sensing and collision avoidance by providing fully-convolutional segmented models capable of per-pixel labeling of an object with a simple 2D image input.
Protoype Faster Than Ever
NVIDIA’s Two Days to a Demo tutorial provides users with all of the necessary tools, including guides, software samples and even pre-trained network models to help you get started building your own deep learning neural network. Whether you are using the Jetson TX2 or a different platform, this tutorial will help you get your own proof-of-concept.
With NVIDIA's step-by-step intro and the power of the Jetson platform, you'll be implementing deep learning into your designs in no time.