This repository provides a pipeline that, given an image classifies clothing parts within a set of attributes, as well as predicts their attributes.
The results are displayed with a web visualization interface.
The different stages of the pipeline are executed as docker containers, so you need to install Docker to run it.
The pipeline uses Docker-Compose (installed by default with Docker) to run the multiple containers. The Docker images are published on DockerHub.
To run the pipeline, execute the following instructions:
- Start the pipeline (all images will be downloaded from DockerHub):
$ docker-compose up
- Open a browser window at http://localhost:5000 (you may need to refresh some times until the first image appears)
The pipeline retrieves the images from the images directory. You can replace the images inside the directory to classify your images.