EyeCloudAI Prepares to Launch a Modular Edge AI Camera Development Kit Range for the Raspberry Pi
Bundled software development kit compatible with Linux on Raspberry Pi, while the hardware also supports OpenVINO.
UPDATE (6/16/2021): EyeCloudAI has now launched its crowdfunding campaign for its Intel Movidius Myriad X-powered edge AI camera development kit, taking orders ahead of a planned late-September shipping date on all hardware.
The company has set pricing at $129 for the Core Board or $149 with the optional 8GB eMMC expansion board for standalone operation. The two megapixel Onsemi AR0234CS global shutter camera sensor is priced at $69, with the eight megapixel rolling shutter sensor priced at $49. For those on a budget, the two megapixel rolling shutter sensor costs just $19.
Orders are now open on the project's Crowd Supply page.
Original article continues below.
Computer vision at-the-edge specialist EyeCloudAI is preparing to launch a camera kit built around Intel's Movidius Myriad X vision processing unit (VPU): the EyeCloudAI CDK.
"EyeCloudAI CDK is an open source edge AI camera development kit for applications where privacy, real-time performance, reliability, and flexibility are paramount," the company explains of its upcoming launch. "With EyeCloudAI CDK, developing and deploying a custom edge AI camera solution easy — we’ve taken care of the AI optimization, sensor tuning, optics, tooling, and supply chain management so you can focus on your application."
The heart of the EyeCloudAI CDK is a "Core Board," built around an Intel Movidius Myriad X MA2085 vision processing unit (VPU) offering 16 vector processors, 20+ vision accelerators, and a claimed 4 TOPS from two neural compute engines, and is designed for compatibility with Intel's OpenVINO Toolkit. This single core board, which includes a USB Type-C connector for host communication and 1GB of LPDDR4 memory, can then be paired with one of three sensor modules.
A two megapixel rolling-shutter sensor is the stock model, designed for general-purpose projects. For those who need higher resolution, this can be upgraded to an eight-megapixel rolling shutter sensor board. Finally, the trio is rounded out with an ON Semi AR0234-based two megapixel global shutter module for high speed work - up to 120 frames per second at one megapixel or 60 frames per second at two, up from the 30 frames per second available on the rolling shutter variants. Finally, an optional eMMC board adds local storage.
"EyeCloudAI CDK works with Intel’s distribution of OpenVINO Toolkit and is compatible with frameworks such as ONNX, TensorFlow, Caffee, MXNet, Kaldi, and PaddlePaddle," EyeCloudAI confirms. "The SDK is compatible with Raspberry Pi flavors of Linux for ARMv7 and ARMv8."
More details, and a link to be alerted when the crowdfunding campaign goes live, can be found on the EyeCloudAI CDK Crowd Supply campaign page; the company has also published schematics for the five boards which make up the kit on GitHub under an unspecified license.