This is a development kit consisting of a VHDPlus Core MAX10 Ultra, VHDPlus IAS Camera Extension and AP-VisiON-AR0830 camera from Appletec.
This sensor has an even higher resolution than the AR0430. The Core MAX10 Ultra has twice the logic elements of the Core MAX10. Together this kit is the ideal entry into FPGA-based ultra-fast neural networks.
This example uses the VHDPlus ultra-low latency and completely hardware-based neural network libraries: https://github.com/leonbeier/OnSemi_CNN_Ultra
You can create your own neural network with TensorFlow and implement it easily with our libraries in the example project.
The CNN for the number detection uses only 25% of the logic elements of the Core MAX10 Ultra, so there is a lot of space for bigger and more advanced neural networks.
Also, the processing speed can be increased a lot. The neural network has almost no delay because everything is calculated while the image is received from the camera. The images per second that can be analyzed are 100 MHz / number of pixels of the first convolution matrix/number of pixels of the image to analyze. For the MNIST number example, a number could be detected 13.000 times per second.
Watch this video to see how the kit works: https://youtu.be/nYiUiKzc0ho
When you use the AI camera kit, select the AR0830 camera as hardware when you create a project. Here the AI demo is an option as well.
The IAS Camera and Display extension allows connecting OnSemi cameras with an IAS connector to an FPGA development board with a CRUVI HS connector. It features an HDMI port that can be used to output the image of the camera. The power supply and oscillator for the camera are also on board, so you just have to connect the extension with the FPGA board and the camera to have a working system. This can be used to program object detection, make your own safety camera, and much more. With the IAS connector with 4 MIPI CSI-2 lanes and a standard HDMI port, you are very flexible with the choice of camera and display you want to connect.
More Information: https://vhdplus.com/docs/components/onsemi_camera