- Clone the repo
git clone https://github.com/erickramer/image_recog.git
cd image_recog
-
Download VGG weights from here and put the weights in the data folder
-
Create virtualenv and install requirements.txt
virtualenv venv
source venv/bin/activate
pip install -r requirements.txt
- Make sure your ~/.keras/keras.json looks like this:
{
"image_dim_ordering": "th",
"epsilon": 1e-07,
"floatx": "float32",
"backend": "theano"
}
- Run app
python run_app.py
- Browse to localhost:5000