An android app that classifies images of food
Uses the AIY model for food classification This model is additionally optimised for inference on ARM based devices using XNNPACK for CPU optimisation and the NNAPI delegate for hardware acceleration if available.
models/ folder
- Has all the code related to the food classification model and other ML stuff
android/ folder
- Has all the Android code
test_images/ folder
- Has test images used to try out the model
aiy_model.tflite
-
The food classification model in the
tfliteformat -
Takes in an input array of
shape = (None, 192, 192, 3)ie(batchsize, height, width, colorchannels)Gives an output ofshape = (1, 1024)ie the score over all 2023 possible food classes
labelmap.csv
- A csv file that maps label numbers to food classes like
3, apple pie