The Kit enables developers to prototype IoT vision applications and facilitates the deployment and (re)-training process of ML-models. The embedded hardware consists of a Basler dart BCON for MIPI camera with a resolution of 13 MP and an NVIDIA's Jetson Nano Developer Board.
The AI Vision Solution Kit enables developers to prototype IoT vision applications and facilitates the deployment and (re)-training process of ML-models, saving time for users setting up an integrated embedded vision system for optimal performance.
Users have access to pre-trained ML models that are available in the cloud and can be deployed on the kit using container technology. Furthermore, users can make use of (re)-training functionalities. Since the quality of the ML inference depends, not only on the training, but also on the processing unit and input data from the camera, Basler ensures that the pre-trained models are optimized for the target hardware of the kit. Once the model is optimized, customers can download the compiled model and deploy it on the target hardware, the NVIDIA Jetson Nano. The ML model deployment process is based on AWS Sage Maker and Sage Maker Neo. While the inference is done on the edge device, resulting meta data is sent to the AWS cloud, using AWS IoT Core.
The kit's embedded hardware consists of the robust and industry-proven Basler dart BCON for MIPI camera with a resolution of 13 MP. The processing board is based on NVIDIA's Jetson Nano, optimized for AI applications due to its strong GPU.
Stay tuned for more information, which becomes available from August 2020.