MobileSam Segmentation Model Service

A FastAPI service developed to deploy the MobileSam segmentation model. The service is containerized with Docker, to ensure easy setup and consistency across different environments. It can handle incoming requests asynchronously which can provide better performance and scalability.

Original Image Output

Description

• FastAPI service defined in app.py exposes the model as a RESTful API.
• The POST endpoint /segment-image: listens to incoming POST requests. When an image file is received, the service:
  • Validates the uploaded file to ensure that it is an image.
  • Reads the file's contents and converts the into byte stream.
  • Opens the byte stream as an image object and coverts it into RGB format.
• Model invocation: The RGB image object is passed to the segment_everything function. This function utilizes the MobileSam model to process the image, performing segmentation, and returns the segmented image.
• Response Formation:
  • The segmented image is converted into a byte array in PNG format.
  • A Response object is created with this byte array as it's content, and the media type is set to image\png.
  • This response, containing the segmented image in binary format, is sent back to the client.

Setting Up and Running the Application

Prerequisites

Before you begin, ensure you have met the following requirements:

• You have installed Docker on your machine. For installation instructions, see Docker documentation

• Add your user to the Docker group:

  sudo usermod -aG docker $USER

Running the application

1. Build your docker image

   docker build -t my-fastapi-app .

2. Run the image as a container:

   docker run -d --name my-fastapi-container -p 8000:8000 my-fastapi-app

3. Access the application

   You can access the application at http://127.0.0.1:8000/docs#/ or http://localhost:8000/docs#/ . Head over to the POST tab, and click on 'Try it out'. Upload an image and Execute. Enjoy your new segmented image.

Testing

1. Automated testing

   To run the automated tests for the API, navigate to the project root and execute

    pytest

   The test scripts are found in the directory test. The images used for testing are the directory test/test_images.

2. Manual Testing with Postman or cURL

• Using cURL:

  curl -X 'POST' 'http://127.0.0.1:8000/segment-image' -H 'accept: application/json' -H 'Content-Type: multipart/form-data' -F 'file=@tests/test_images/valid_image.jpg;type=image/jpeg' --output segmented_image_cURL.png
  curl -X 'POST' 'http://127.0.0.1:8000/segment-image' -H 'accept: application/json' -H 'Content-Type: multipart/form-data' -F 'file=@tests/test_images/invalid_file.txt'

• Using Postman

  • Open Postman and create a new request
  • Choose POST as the request method
  • Set the URL to http://localhost:8000/segment-image
  • Under the Headers tab, set Content-Type to multipart/form-data.
  • Under the Body tab,
    • select the form-data option.
    • In the key field, enter file
    • Set the type to File from the dropdown on the right.
    • Use the Choose Files button to select an image file from your system; or from the images which can be found in the test/test_images directory.
  • Send the Request
  • Review the Response

Footnotes

• Fast API service, Docker setup and testing scripts built by Davis Joseph