It's a simple application that allows you to upload an image and performs classification on it.
First, user clicks upload and select an image:
Then the server will run the image through a residual neural network with 50 layers and compute an answer:
The easiest way to run this project is through docker. First, build the image.
docker build -t tensorgo .
And then run the image in a container and publish port to 3000
docker run --publish 3000:3000 tensorgo
Now you can see the app on localhost:3000!
If you are experiencing error regarding model failed to load or complaint from TensorFlow about
cannot find model, then that means you probably need to run the Python script to build the model
first and then build Docker image.
Create a virtual environment and install requirements using pip:
virtualenv environment
source environment/bin/activate
pip install -r requirements.txt
Now navigate to tf_models/ folder and run one of the Python scripts:
python create_resnet_model.py
Now build again:
docker build -t tensorgo .
- Go 1.8+
- Node 6+
- Python 2.7+
Source: Execute the following shell command
TF_TYPE="cpu" # Change to "gpu" for GPU support
OS="linux" # Change to "darwin" for macOS
TARGET_DIRECTORY="/usr/local"
curl -L \
"https://storage.googleapis.com/tensorflow/libtensorflow/libtensorflow-${TF_TYPE}-${OS}-x86_64-1.8.0.tar.gz" |
sudo tar -C $TARGET_DIRECTORY -xz
Then configure the linker
sudo ldconfig
Once TensorFlow is installed, next is to install Go dependency
dep ensure
Then compile Go source code
go install
Install node modules for building the frontend
npm install
Then build it
npm run build
Create a virtual environment and install requirements using pip:
virtualenv environment
source environment/bin/activate
pip install -r requirements.txt
Now navigate to tf_models/ folder and run one of the Python scripts:
python create_resnet_model.py
Now you are ready to go.
Simply do
tensorgo start