AI Service Microservice

This project contains two main components:

1. UI Backend: A Flask-based service that handles image input from the user, stores the image, and interacts with the AI backend for object detection.
2. AI Backend: A Flask-based service that processes the image using a lightweight object detection model (e.g., YOLOv5).

The UI communicates with the AI backend to perform object detection on images and returns results in a structured JSON format.

Prerequisites

Before running this solution, ensure you have the following tools installed:

• Python (Flask or FastAPI): For building the backend services.
• Docker: For containerizing the application (optional).
• Lightweight Object Detection Model: An open-source model such as YOLOv5 or MobileNet SSD (we recommend YOLOv5 for object detection).
• Python Libraries:
  • Flask: For Python based backend server.
  • requests: For HTTP communication between the UI and AI backend.
  • Pillow: For reading the PIL image
  • torch: For accessing YOLOv5 model from Hub
  • opencv-python: For drawing bounding box into output image

Architecture Overview

The system consists of two services:

1. UI Backend:

   • Accepts image input from users.
   • Stores the image and calls the AI backend to process it.
   • Serves the processed image with bounding boxes and detection results in JSON format.
   • Port: 5000

2. AI Backend:

   • Accepts image input from the UI backend.
   • Uses an object detection model (e.g., YOLOv5) to process the image.
   • Returns the bounding boxes and detected objects in a structured JSON format.
   • Port: 5001

Environment Variables

In the UI service, ensure the following environment variable is set to point to the AI backend:

AI_BACKEND_URL=http://ip-address:5001

Running the Services

1. Running Without Docker

UI Backend

• Navigate to the ui directory.
• Install the necessary dependencies:

pip install -r requirements.txt

• Run the UI backend:

python app.py

• The UI should be accessible at http://127.0.0.1:5000.

AI Backend

• Navigate to the ai-backend directory.
• Install the necessary dependencies:

pip install -r requirements.txt

• Run the AI backend:

python app.py

• The AI backend should be accessible at http://127.0.0.1:5001.

2. Running with Docker and Docker Compose

Prerequisites

• Ensure Docker is installed on your machine.
• Ensure Docker Compose is installed. Steps
• Clone the repository to your local machine:

git clone https://github.com/AmitAaranya/ai-service.git

• Navigate to the root directory of the project:
• Build and run the services using Docker Compose:

docker-compose up --build

• The UI should be accessible at http://127.0.0.1:5000.
• The AI backend will be running at http://127.0.0.1:5001.

Interacting with the Services

Upload an Image:

• Access the UI via the browser at http://127.0.0.1:5000.
• Upload an image for processing.

Processing by AI:

• The UI backend sends the image to the AI backend for object detection.
• The AI backend processes the image, applies the object detection model, and returns the results.

Output:

• The UI backend returns the processed image with bounding boxes and a structured JSON response containing the detected objects.

{
    "detections": [
        {
            "confidence": 0.889124870300293,
            "coordinates": {
                "height": 255.1783905029297,
                "width": 137.53494262695312,
                "x": 172.6475830078125,
                "y": 298.4649963378906
            },
            "label": "dog"
        },
        {
            "confidence": 0.6902128458023071,
            "coordinates": {
                "height": 77.86497497558594,
                "width": 169.74417114257812,
                "x": 453.25555419921875,
                "y": 96.6738052368164
            },
            "label": "car"
        }
    ]
}

References

• Loading pre-trained YOLOv5 model via PyTorch Hub
• Getting start with Docker
• Docker Compose documentation
• Python Flask documentation