I am working on image classification using OpenMV. One of the labels I am working on is human legs. In the live classification, the model works perfectly to classify images containing human legs. But once deployed on the OpenMV M7 Camera, the model does not recognize it correctly at all. Any advice on why this is happening, and how to correct the model so that it works in real time as well would help. The project Id number is 25777
I noticed your dataset uses 60x60px images, are you also cropping images to this format on the OpenMV?
What I would suggest to do to pinpoint the issue: collect images from the OpenMV IDE and add them to your test set. You can follow this guide: https://docs.edgeimpulse.com/docs/image-classification-openmv
Then you can verify in live classification/model testing if classification is correct for those new samples. If it works fine then it’s probably an issue in the OpenMV python script.