Thanks! 

Hi Megan,

The Sony IMX500 (Raspberry PI AI camera) can only process a maximum image size of 640x480 pixels, which is pretty small. This means the object target has to be relatively large in the image to be discernible. The example video at the end of that youtube video is a good indication of what the size in the image will need to be. Though the level of detail required (aka size in the image etc) for correct species identification will of course depend on species.

If you want to process higher resolution images so you can detect things that are smaller in the image (small objects, or things far away) you can use something like the Hailo 8 accelerator with the Raspberry PI 5, I talk about using it in these videos:

As for speed of movement, that will depend on inference speed (how many images you can process per second). The IMX500 can process frames at 30 fps. The Hailo 8 can go faster, but that will depend on the model size, image size, and camera capabilities. 

Let me know if you have any further questions!

Regards

Luke