How do we implement the FOMO algorithm into Python?

Help
#1

I am unsure about how to port the FOMO algorithm into Python. I tried using the downloaded TFLite files in the dashboard, but I get dimension errors such as “ValueError: Cannot set tensor: Dimension mismatch. Got 2 but expected 4 for input 0.”. Additionally, I used the Python for Linux SDK, but when I run the classify.py, I get a “Exec format error” for the eim file. I am trying to do real-time object detection.

Thank you!

#2

Hi @olfk.1234,

I recommend importing the .tflite file into netron to see what the expected input dimensions should be. It looks like you’re trying to feed a 2D array to the FOMO model, but the model expects a 4D input (sample, width, height, channel). See here for more info.