FOMO model Heatmaps

[Netra13]

I’m trying to use the FOMO model file locally to output a ‘heatmap’ containing classified images for post processing, is there a way to do that by editing any code in the existing model file that I can grab from EI?
I am trying to run a classification model on a cortex M4F (nano BLE 33) and thus am able to only fit a lightweight model like FOMO on it to detect empty spaces in a shelf ~ in a retail setup. It shows me the count and location but I wanted to try and get any data on the dimensions of the empty space as well through my image classifier (maybe through post processing).

Also, would it be a feasible solution if I ran a regression classifier instead of an object detection classifier by labelling the % empty space for each image and training it on EI?

[louis]

Hello @Netra13,

That thread might interest you:

FOMO model and heatmap to centroid

Hi @matkelcey Thanks for the information, I’ll definitly have a look in the code you mentioned and try to implement the thresholding and fusing in my own script! And indeed, I also have the feeling that the exported model does not equally behave like the one that is running online… I don’t think it is transposed because when there is a totally different image shown, the image can be recognised in the heatmap and follows the same directions when moving the image. Does it make sense that you re…

Best,

Louis