Today at GPU Technology Conference(GTC) 2020, NVIDIA announced a new 2GB Nvidia Jetson Nano for the first time. Last year, during the March timeframe, NVIDIA introduced the $99 Jetson Nano Developer Kit which came with 4GB 64-bit LPDDR4 Memory. The new NVIDIA Jetson Nano 2GB Developer Kit, priced at $59, makes it even more affordable for students, educators, and enthusiasts to learn AI and robotics.
Pre-order Now: https://nvda.ws/30v5w3M
The NVIDIA Jetson Nano 2GB Developer Kit is the ideal platform for teaching, learning, and developing AI and robotics applications. It uses the same proven NVIDIA JetPack Software Development Kit (SDK) used in breakthrough AI-based products. The new developer kit is unique in its ability to utilise the entire NVIDIA CUDA-X™ accelerated computing software stack including TensorRT for fast and efficient AI inference -- all in a small form factor and at a significantly lower price. The Jetson Nano 2GB Developer Kit is priced at $59 and will be available for purchase starting end-October.
JetPack SDK & Libraries for AI development
With Jetson Nano 2GB, the JetPack SDK comes pre-loaded with all the necessary libraries one would require to build AI applications. For example, OpenCV and Visionworks for computer vision and image processing, CUDA, cuDNN, and TensorRt to accelerate AI inferencing, libraries for camera and sensor processing, and much more. Getting your Jetson Nano 2GB up and running with the JetPack SDK takes only a few steps.
Popular Frameworks shipped as Docker containers
Jetson Nano 2GB enables you to learn and develop using the framework of your choice by supporting all popular frameworks including TensorFlow, PyTorch, and MxNet. Development containers for TensorFlow and Pytorch are hosted on NVIDIA NGC which provides a quick one-step method to get your framework environment up and running. The familiar Jupyter Notebook learning environment is also available on Jetson Nano 2GB using the machine learning container hosted on NGC.
Jetson Nano 2GB Developer kit is capable of running neural network training on-device. Of course, training a neural network from scratch requires a lot of compute resources and is not practical to run on Jetson Nano 2GB. But, one can easily run Transfer-Learning training jobs locally on the Jetson platform. The Jetson Deep Learning Institute (DLI) course illustrates the simplicity of the process by teaching you to modify a trained network through transfer-learning. For robotics learners, the JetBot Robotics Kit illustrates the process to train a neural network that can be used by the robot for collision avoidance and following road markings.
80,000+ Community members and still counting…
The Jetson community consists of 80,000+ active members that includes AI enthusiasts, researchers, developers, students, and hobbyists who are bonded by their passion for AI and the Jetson platform.Jetson community members create cool projects, applications, and demos that are shared with the community and best of all, they share the code and instructions for anyone to try, learn, and enhance these projects.
Check out Jetson community projects built for Jetson Nano 2GB Developer Kit
Let's look at the specifications of the latest NVIDIA Jetson Nano 2GB Developer kit.
Is $59 Jetson Nano a Raspberry Pi Killer?
Last year, Raspberry Pi 4 was announced priced at $35 with 4K support and up to 4GB of RAM. It came with up to 4GB of RAM (four times that of any previous Pi), dual-band Wi-Fi, twice the amount of HDMI outputs, and two USB 3 ports. Below is the comparison chart which NVIDIA published comparing Jetson Nano 2GB Developer Kit features with Raspberry Pi 4 and Google Coral Development board.
Check out Jetson community projects built for Jetson Nano 2GB Developer Kit in the below video:
Compared to Raspberry Pi 4 and other development kits available at similar price points, the Jetson Nano 2GB not only supports all the popular AI frameworks and networks, but also delivers orders of magnitude higher AI performance.
The chart below shows the AI inferencing performance of Jetson Nano 2GB on popular DNN models for image classification, object detection, pose estimation, segmentation, and others. The benchmark was run with FP16 precision using JetPack 4.4.
The world of AI computing is changing fast. Researchers are constantly inventing new neural network architectures that deliver better accuracy and performance for certain tasks. The wide variety of AI models and frameworks in use is evident by examining the various projects found in the Jetson community projects portal, some of which are highlighted in the Community Projects section of this document. This new 2GB Jetson Nano is surely an affordable device for students to learn and create AI projects should therefore be flexible enough to run a diverse set of AI models and also deliver performance required to create meaningful interactive AI experiences.