Image Recognition App

Image Recognition App is an image classification application using Django and PyTorch.

Requirements

• OS: Windows10 Pro (It will probably also work with Linux, but has not been tested yet.)
• CPU only (not support GPU)
• Python 3.8

Installation

$ git clone https://github.com/noriho137/image-classifier.git
$ cd image-classifier/
$ python -m venv {venv_name}
$ {venv_name}/Scripts/activate.bat
({venv_name}) $ pip install -r requirements.txt

Download pretrained model and label definition

Pretrained model

This application use pytorch pretrained model. For example, following models are available.

Name	URL
alexnet	https://download.pytorch.org/models/alexnet-owt-4df8aa71.pth
densenet201	https://download.pytorch.org/models/densenet201-c1103571.pth
googlenet	https://download.pytorch.org/models/googlenet-1378be20.pth
inception_v3	https://download.pytorch.org/models/inception_v3_google-1a9a5a14.pth
mobilenet_v2	https://download.pytorch.org/models/mobilenet_v2-b0353104.pth
resnet50	https://download.pytorch.org/models/resnet50-19c8e357.pth
squeezenet1_0	https://download.pytorch.org/models/squeezenet1_0-a815701f.pt
vgg16	https://download.pytorch.org/models/vgg16-397923af.pth

Label definition

This model need class label definition imagenet_class_index.json from https://s3.amazonaws.com/deep-learning-models/image-models/imagenet_class_index.json .

Generate Django secret key

Generate Django secret key by secrets.token_urlsafe.

({venv_name})$ python -c "import secrets; print(secrets.token_urlsafe(38))"

Set up environment variables

Make .env file at project root directory and define environment variables like following:

# Django settings
export DJANGO_SECRET_KEY={Django secret key generated above}
export DJANGO_DEBUG=True

# Classification model
export PRETRAINED_MODEL_NAME=vgg16
export PRETRAINED_MODEL_PATH=./pretrained/vgg16-397923af.pth
export CLASS_INDEX=./pretrained/imagenet_class_index.json
export MAX_RANK=5

How to run

({venv_name})$ python manage.py migrate
({venv_name})$ python manage.py runserver

Browse to http://127.0.0.1:8000/ .