EyeCloudAI CDK

An open edge-AI camera kit compatible with OpenVINO and based on the Intel Myriad X

Limited items in stock

View Purchasing Options
Jun 24, 2021

Project update 2 of 10

The Story Behind the Three ISP-ready Sensor Modules To Be Available with EyeCloudAI CDK

by Johanna S

Our story started with an attempt at making a home security AI camera, which was designed for that single purpose. When we failed several times to integrate it into other systems or products, we realized an open AI vision solution was needed. This led to the idea of OpenNCC (Open Neural Computing Camera). We knew it should be flexible, easy to integrate, and contain embedded AI functions.

In 2019, this idea was turned into a flexible education kit that can be used as hands-on tools in AI vision teaching and which also can be integrated into other STEAM courses and devices.

Later that year, the first version of the kit came out, which combined one core board based on the Intel Movidius VPU with a 2 MP sensor module. We designed it as 38 mm x 38 mm because one of our customers is from the manufacturing industry, and 38 mm x 38 mm is a widely used standard in that segment. In this version, we solved customer’s headaches of transferring AI models to the boards by integrating with OpenVINO and offering an SDK. Customers can train their models in various frameworks, then use the OpenVINO toolkit and our SDK to deploy the models to the AI camera kit. Additionally, to meet higher resolution requirements, we added a 38 mm x 38 mm 8MP sensor module to the kit.

In 2020, the pandemic came. We worked quickly with our partner and equipped the kit with an infrared module to produce a fever screener with mask detection functions to help the community.

When developing for a fast moving subject application, we found a normal rolling shutter sensor did not work because of the distortion in fast-moving scenarios. So we developed the AR0234 sensor module, also in 38 mm x 38 mm. Below you can see a prototype of the OCR recognition system being used for scooter frame rolling code recognition.

For this campaign, we set the goal at $1 because we will produce EyeCloudAI no matter what to work on projects with our customers and partners. The purpose of this campaign is to find out whether there are other developers who need the kind of solution we did. If you do, please back us and make it a successful campaign.

Sign up to receive future updates for EyeCloudAI CDK.

Subscribe to the Crowd Supply newsletter, highlighting the latest creators and projects