The Google Coral M.2 Accelerator with Dual Edge TPU is a versatile M.2 module designed to enhance existing systems and products with an M.2 E-key slot. Each of the two Edge TPU coprocessors can perform an impressive 4 trillion operations per second (4 TOPS) while using only 2 watts of power, enabling efficient execution of mobile vision models such as MobileNet v2 at nearly 400 FPS. Utilising both Edge TPUs in the module allows for doubling inferences per second (8 TOPS) in various ways, including running two models simultaneously or pipelining a single model across both coprocessors. This accelerator is compatible with Debian-based Linux or Windows 10 systems and works seamlessly with TensorFlow Lite models, eliminating the need to build models from scratch.
Disclosure: This post contains affiliate links. If you click through and make a purchase, I will earn a commission, at no additional cost to you. Read my full disclosure here.