ALPR on Nvidia Jetson Devices

What are Nvidia Jetson Devices?

Nvidia Jetson devices are embedded computing boards made by Nvidia. They are computing platforms that are designed for a specific software-controlled task mostly to accelerate machine learning applications like deep learning.  Jetson devices are generally designed to be power efficient.  The boards typically have a processor, a memory unit, an input/output peripheral device that has a dedicated function.

There are many variants of Jetson devices ranging from Jetson Nano, Jetson TX1, Jetson TX2 to Jetson Xavier.  Each Jetson devices has its unique functionality and improved capacity for different AI applications.

• Jetson Nano is a small and powerful computer for embedded applications that delivers the power of modern AI.  It was designed mainly for hobbyist robotics due to its low price. The final specification show that the board is a sort of power-optimized, stripped down version of a full Tegra X1 system.
• Jetson TX1 is the first set of supercomputers designed on a module. It has the ability to produce the performance and power efficiency needed for the visual computing applications. It is built around the revolutionary NVIDIA Maxwell™ architecture with 256 CUDA cores delivering over 1 Teraflops of performance.  It is a very good system for embedded deep learning, computer vision, graphics, and of course ALPR.
• Jetson TX2 is the fastest and most power-efficient embedded AI computing device. This is one of the best supercomputer modules you can get that brings true AI computing at the edge with a power consumption of only 7.5 Watt.  It is built around an NVIDIA Pascal™-family GPU and loaded with 8GB of memory and 59.7GB/s of memory bandwidth. It has a lot of standard hardware interfaces that makes it easy to integrate into a wide range of products and form factors.
• Jetson Xavier NX is very efficient in running modern AI workloads and consumes as little as 10 watts of power.  It has a compact form factor smaller than a credit card. Jetson Xavier runs modern neural networks in parallel and process data from multiple high-resolution sensors creating a platform for embedded and edge computing devices that demand increased performance but are constrained by size, weight, and power budgets.
• Jetson AGX Xavier is the most powerful embedded system-on-module with respect to other Jetson modules and has the capability of a desktop GPU workstation while operating with just 10 Watts of power.  Jetson AGX Xavier makes neural networks easy to train and implement quickly and also opens the door for many AI-powered autonomous machines to thrive.  It has high computing performance with 32 TeraOPS.