It’s a beyond doubt, Artificial Intelligence (AI) is going to change or changing the way we live, play and interact with things. From serving as a virtual assistant to auto completing our search queries, AI offers an immense potential to make our lives a lot more comfortable. However, these sophisticated AI algorithms require an enormous processing hardware power for training the modules and for deployment. Despite providing huge benefits this has tend to reduce the use of AI due to power and cost constraints. NVIDIA tries to close this gap by introducing a powerful hardware on its Jetson Nano that can support training models for high end AI applications at an affordable cost of <100$ by bringing AI to DIY.
By this time there would be tones of posts talking about the power of Jetson Nano Hardware. But the main aim of this post is to point you to rich set of open source resources and community forums available for you to get started with AI, kindle your passion to dive into this ocean of AI by providing pointers to the innovation happening and practical applications being developed out of it.
Jetson Nano is a small powerful computer for AI and embedded applications that delivers the power of modern AI in 99$ module (59$ for a 2GB version). It hosts a powerful Quad-core ARM CPU and NVIDIA Maxwell Architecture GPU with 128 CUDA Cores with 4GB RAM. Talking about interfaces, it supports USB 2.0 and 3.0 ports, plus multiple GPIOs and supports SPI, I2C protocols. It provides a fast network connectivity with Gigabit Ethernet and Wi-Fi 802.11 ac.
Like a racing car needs a good driver, to harness the power of this hardware, we need a well-defined, well supported and continuously upgraded software stack. With AI applications, basically a software defined system, this has become more important than ever. Besides the clever GPU-accelerated processor technology, NVIDIA also provides a comprehensive software package. Unified GPU Architectures enables the software tools and libraries that run on a multi-million-dollar hardware to run on the 59$ priced Jetson Nano 2GB. This unique capability allows Jetson Nano to develop and deploy high-end AI applications and makes the source code portable (reusable) when you decide to upgrade your hardware. And the support for cloud-native technologies like containerization and orchestration makes it easier to build, deploy, and manage AI at the edge.
JetPack SDK provides an ideal platform to start learning AI, it provides more flexibility by making available the development SDK with source. Jetpack SDK includes the latest Linux Driver Package (L4T) with Linux operating system and CUDA-X accelerated libraries and APIs for Deep Learning, Computer Vision, Accelerated Computing and Multimedia, and supports high level SDKs for streaming video analytics and robotics.
Sites to learn more on jetpack and installation:
- Jetson Developer Forum
This great developer zone helps you get the most out of every project with an active, knowledgeable, creative community.
- Jetson AI Certification Program
Nvidia with its Deep Learning Institute offers hands on training on AI, accelerated computing and accelerated data science. If you already own a Jetson Nano, then there are free courses available to you. There are other courses on TensorFlow, High Performance Computing that you would be interested in.
- Open Source Projects
There are a tones of open source AI projects available with videos and code for you to get started with the Jetson Nano. And this developer forum offers exciting gifts to developers each month by choosing a best project submitted. Check out some of the cool projects available at https://developer.nvidia.com/embedded/community/jetson-projects