Hi，Is it possible to realize image recognition on the hardware of openmv H7 without plus?
It will work, you may need to choose a smaller MobileNet model (0.1 or 0.05) and convert to grayscale. Here’s a good tutorial on running image recognition on the H7: https://www.hackster.io/ishotjr/mega-or-mini-image-classification-on-the-1mb-openmv-cam-h7-be57ac
And make sure to make your image size smaller, e.g. 48x48 on the Create impulse screen.
thank you, ,i will try it
Thank you very much. I’ll try it.
ok, thank you，I will pay attention to that
I have to build an object recognition application. The goal is to detect the nature of an object given an image (metal, cartoon, glass, paper, plastic).
I have a openmv H7 (without plus).
I got a labeled dataset of labeled pictures. There are 5 classes (metal, cartoon, glass, paper, plastic).
Is it possible to import this labeled dataset in Edge impulse ?
More generally is it possible to build this object recognition application with the openmv H7 thanks to Edge Impulse ? Can you share some resource(s) to achieve this task ?
You should be able to run your model on the OpenMV H7, please check out this thread: OpenMV H7 Memory Error:Out of Fast Frame Buffer Stack Memory
Also, to upload your labeled image on the studio, you can use either the online uploader or the cli (
edge-impulse-uploader). Please see our documentation on this topic: https://docs.edgeimpulse.com/docs/cli-uploader
Thank you for your reply !
I will try your suggestions…
Excellent ! I did deploy my model on the device : It perfectly worked !
Now I have to find the best model…
Once again thank you for your advices !