Coral System on Module (SoM) is a fully-integrated Linux system for accelerated Machine Learning (ML) inferencing to be integrated into existing hardware with three 100-pin connectors. The SoM contains the NXP iMX8M SoC, eMMC memory, LPDDR4 RAM, Wi-Fi, Bluetooth, and the Google Edge TPU Coprocessor for acceleration.
The Edge TPU Coprocessor can execute state-of-the-art mobile vision models such as MobileNet v2 at 400 FPS in a power efficient manner. It is capable of performing 4 trillion operations (tera-operations) per second (TOPS), using 0.5 watts for each TOPS (2 TOPS per watt). This on-device processing reduces latency, increases data privacy, and removes the need for a high-bandwidth connection for cloud performance. Models are built using TensorFlow. Now available in 1GB, 2GB, and 4GB LPDDR4 SDRAM versions.
Coral, a division of Google, helps build intelligent ideas with a platform for local AI.