T-DEV-810

AI: Image recognition

Using Virtualenv

Install virtual env

pip install virtualenv

cd ./T-DEV-810/

Init virtualenv

virtualenv -p python3.7 venv3.7

Use existing virtualenv

Ubuntu : source venv3.7/bin/activate

Windows: venv3.7\Scripts\activate.bat

Install package in virtualenv

pip install <module>

Save dependencies

pip freeze > requirements3.7.txt

Load dependencies

pip install -r requirements3.7.txt

Stop using virtualenv

deactivate

Jupyter notebook

jupyter only works with python2.7

cd ./T-DEV-810/

pip install virtualenv

virtualenv -p python2.7 venv2.7

source venv2.7/bin/activate : to activate virtual environment

pip install -r requirements2.7.txt

jupyter notebook notebook-1.ipynb

deactivate : deactivate virtual environment

Tensorboard

Activate logs

Edit init_or_load_then_train_main.py : CnnModel(filename, inputShape, 3, False) => CnnModel(filename, inputShape, 3, True)

Train with activate logs

python init_or_load_then_train_main.py ${model_log_folder}

Launch tensorboard server

tensorboard --logdir logs/fit : default:: /localhost:6006/