A simple logo classifier developed using Maixduino framework and PlatfomIO, to run on K210 MCU on Sipeed's Maix dev board.
A simple logo classifier developed using Maixduino framework and PlatfomIO, to run on K210 MCU on Sipeed's Maix dev board. I trained my own ML model, using transfer learning from MobileNet v1.
Click the thumbnail
nncase
for NeuralNet optimization, download from here. Unzip anywhere.mobilenet.py
file on site-packages/keras_applications
(don't forget to backup) with the one in this repo. site-packages
folder may exist on several places depends on your environment. If you use virtualenv, it should be under you_virtualenv_dir/lib/python3.x
training/mbnet_keras.py
file. Adjust the constants, and run it.h5
model file by running training/convert.sh
script with the h5 model file as parameter. Eg. ./convert.sh logoclassifier.h5
src
src/names.cpp
file(More complete steps will be coming soon)