Hotdog Not Hotdog image classifier with TensorFlowJS. Demo link
Generated using [email protected]
This project uses Gulp to run automated tasks for development and production builds. The tasks are as follows:
gulp --production
: Same as gulp serve --production
also run gulp test
and not boot up production server
gulp serve
: Compiles preprocessors and boots up development server
gulp serve --open
: Same as gulp serve
but will also open up site/app in your default browser
gulp serve --production
: Same as gulp serve
but will run all production tasks so you can view the site/app in it's final optimized form
gulp test
: Lints all *.js
file in the source
folder using eslint and runs all *.test.js
file unit tests through Karma and Jasmine
gulp test --watch
: Same as gulp test
but will constantly watch *.test.js
files and rerun tests when changes are detected
Adding the --debug
option to any gulp task displays extra debugging information (ex. data being loaded into your templates)
- Docker (https://store.docker.com/search?type=edition&offering=community): for initial setup of machine learning bits
training
├── Docker
└── data
└── images
├── hotdog
│ ├── hotdog1.jpg
│ ├── hotdog2.jpg
│ ├── hotdog3.jpg
│ └── ...
└── not-hotdog
├── not-hotdog1.jpg
├── not-hotdog2.jpg
├── not-hotdog3.jpg
└── ...
Go to the root folder of the project and run the following code with Docker running in the background. Change /path/to/data
to the full path of where the training images are. Take note that folders must not have spaces. It will cause training issues. The Docker file is included at the root folder of this project.
$ cd training
$ docker build -t model-builder .
$ docker run -v C:/xampp/htdocs/not-hotdog-pwa/ML/training/data:/data -it model-builder