I’m still somewhat of a novice when it comes to this stuff, but I’m doing my best trying to learn. I’ve already installed and tested the detection software on my Pi, now I wanna try to run it with a custom dataset instead of the default Coco Library.
I’ve seen Tim recommend using Edge Impulse to build a custom dataset, which I’ve already done. Now my problem is, how do I change out the Coco Library with the dataset that I’ve made? Can anyone here help me out with this?
I think you can try Tensorflow instead.
Hey Andrew.
The fastest way is to find the directories in your version of Raspberry Pi OS which has these three files in them. Those three files make up the default Coco library.
Edge impulse should have created for you similar files likely named other things. Find the matching files and (to save you from altering the Python Scripts) save over those original files with the ones produced by Edge Impulse with the exact same name. Naturally matching the file type to the corresponding file type as you do so.
With that done you should be golden 
.
Best of luck champ!
Tim
Sorry for the late reply. Thanks for answering, Tim! I’ll try and give Edge another go. I’ll reply back here if I manage to succeed or not.
And you can get our latest projects and tips straight away by following us on:
