This service uses pre-trained models to detect objects in images with a set accuracy percentage. We prepared a docker container preloaded with all the necessary libraries to run this service.
- Install Docker https://docs.docker.com/get-docker/
- Clone this repository
git clone [email protected]:latam-03-at/ml-service.git- Go to the repository directory that you just clone
cd ml-service
- Build the ml-service container
docker-compose up -d
- You are now able to use the service.
We use Postman for this service. To configure the collection:
- Create a POST request.
- Enter as request URL the following: http://localhost:3000/api/v1/recognize-objects
- Set the following variables:
- percentage (you can set any value from 0.00 to 1.00. E.g. "0.7")
- object (you can write the object required. E.g. "dog")
- model (you can write the pre-trained model that you want to use. You can choose between "yolo", "coco" or "mobilenet". E.g. "coco")
- images (You can upload a single image or an array of images that you want to analyze in jpg or png format)
- Click on "send".