MGL869-IA-project

The NodeJS MVC application

How to run the web-server

1. Run npm install in the root folder.
2. Build the docker using the docker-compose.yml file.
3. The frontend should be available at http://localhost:3000/.

The observability stack

• Prometheus: http://localhost:9090/
• Grafana: http://localhost:3010/
• Kibana: https://localhost:5601/ Username=elastic / Password=changeme

The fastAPIs

To build the fastAPIs docker images, you should first build the base image with the fastAPI.base-image/Dockerfile file. You can use the following command from the folder fastAPI.base-image:

docker build -t fastapi-base-image .

Then, you can build the fastAPIs using the docker-compose.yml file.

Data processing

In order to work with the data processing fastAPI, you should first create two buckets in your MinIO instance http://127.0.0.1:9001/browser, by default, the buckets are images and training-data. User is minioadmin/minioadmin.

Then, you have to put the images bucket in download status (read access for all users)

mc alias set local http://localhost:9000 minioadmin minioadmin
mc admin policy create local readonly-policy minio-policy.json 
mc anonymous set download local/images 

Then start the api. If everything is ok, you should be able to access the api at http://localhost:5010/docs and a table name images should have been created in the postgres database.

After that, you will need to fullfill the buckets with sample images to be able to test the pipeline. To do that, just use the image of the dataset contains in seg_pred and add them to a folder named images under folder setup_image. Then you can run init-db-minio which is in the same folder. That will copy images in the minio bucket and create entries in the postgreSQL BD with their categories.

After that, wait for a while that all the images are loaded in the bucket. You can check the images by going to http://localhost:9001/ and logging in with the credentials minioadmin:minioadmin (defaults).

After doing so, you can run the pipeline directly from the data processing fastAPI. The output images will appear in the other bucket named training-data (default).