For this project, we are actually going to create two different projects. We will break up our requirements for each project separately.

First, we have standard object detection. For this project, our software requirements are as follows:

• Use the CIFAR10 classes to detect common objects.
• When an object is detected, use the OpenMV Cam H7 GPIO to turn on an LED.

Second, we have person detection. For this project, our software requirement is as follows:

• When a person is detected, use the OpenMV Cam H7 GPIO to turn on an LED.

We could list the specific frameworks that we want to use, such as the CMSIS-NN, but again, as we develop requirements, we don't necessarily want to design the system for the developer or constrain their ability to solve the problem unless it is absolutely necessary. For this project, we could also add a number of interesting requirements, such as writing a recognition log to an SD card, communicating a data packet back to a master processor, or something more complex. I'm going to leave those interesting additions to you though once you understand how to detect an object.

Let's now dig into the design for our system and discuss the software theory on how we can recognize objects.