Google Coral SBC was the first building board with Google Edge TPU. The AI accelerator was mixed with an NXP i.MX 8M quad-core Arm Cortex-A53 processor and 1GB RAM to provide an all-in-all AI edge computing platform.
It launched for $175, and now quiet retails for $160 that would not be cheap to students and hobbyists. Google launched a new mannequin called Coral Dev Board Mini final January, and the correct news is that the board is now on hand for pre-snarl for perfect below $100 on Seeed Studio with shipping scheduled to delivery by the head of the month.
Coral Dev Board Mini specifications haven’t modified noteworthy for the rationale that usual announcement, but we all know a few more most important facets:
- SoC – MediaTek MT8167S quad-core Arm Cortex-A35 processor @ 1.3 GHz with Imagination PowerVR GE8300 GPU
- AI/ML accelerator – Google Edge TPU coprocessor with up to 4 TOPS as fragment of Coral Accelerator Module
- Machine Memory – 2GB LPDDR3 RAM
- Storage – 8GB eMMC flash reminiscence
- Connectivity – 802.11a/b/g/n/ac Wi-Fi 5 and Bluetooth 5.0 through MediaTek MT7668 wireless chip
- Video Output
- micro HDMI 1.4 port
- 39-pin FFC connector for 4-lane MIPI-DSI expose
- Video – 720p video encoding/decoding
- Digital camera – 24-pin FFC connector for 4-lane MIPI-CSI2 camera
- Audio – 3.5mm audio jack; digital PDM microphone; 2.54mm 2-pin speaker terminal;
- USB – 2x USB 2.0 Sort-C ports
- Growth – 40-pin GPIO header
- Energy Provide – By capability of USB-C port; MT6392 PMIC
- Dimensions – 64 x 48 mm
The new board is more compact and affords double the reminiscence, but comes with a slower processor which will or may maybe maybe well maybe also merely not influence AI inference efficiency, and lacks Ethernet, stout-sized USB ports, and a constructed-in microphone. While the usual board had an HDMI 2.0 port, the Mini board comes with a micro HDMI 1.4 video output as an more than a few.
The enlighten cases and capabilities of Coral Dev Board Mini single-board pc remain the equivalent with the platform designed for machine learning (ML) inferencing the utilization of the Edge TPU coprocessor faithful of performing 4 TOPS the utilization of 2 TOPS per watt in theory. As an instance, it will speed MobileNet v2 at shut to 400 FPS at low vitality the utilization of 224×224 pictures somewhat than the more worn 300×300 pictures as we reported in our put up about determining AI benchmarks.
The board runs Debian primarily based entirely Mendel Linux distribution developed by Google for Coral boards and supports TensorFlow Lite and AutoML Imaginative and prescient Edge with the latter enabling “snappy, high-accuracy personalized image classification objects”.