CloudRaft Assignment - Image Classification Service

This project is an implementation of the CloudRaft assignment for creating a backend application with a pre-trained model for image classification. The application allows users to upload images, classify them into categories (simulating dog breed classification), and query the uploaded images.

video

Video

Demo

Live Demo

Postman API Documentation

Documentation

Endpoints

GET /model

Retrieves information about the model.

Response:

{
  "name": "MobileNet (simulating dog breed classifier)",
  "version": "1.0",
  "description": "Pre-trained model for image classification"
}

POST /img
- Accepts an image and returns its classification.
- Request: Form-data with key "image" and file upload.
- Response:
```
{
  "category": "labrador",
  "uid": "1519151980"
}
```
GET /images
- Returns all uploaded images with their UIDs and categories.
- Response:
```
[
  {
    "uid": "6569841",
    "category": "labrador" 
  },
]
```
POST /img/{uid}
- Downloads the image that matches the provided UID.
- Response: Image file download.
GET /categories
- Returns all categories with the number of images in each.
- Response:
```
{
  "labrador": "2",
  "bulldog": "12"
}
```

Local Setup

Prerequisites

Node.js (v18.17.0 or later)
npm (comes with Node.js)

Installation

Clone the repository:

git clone https://github.com/sagnik3788/CloudRaft-assignment.git
cd CloudRaft-assignment

Install dependencies:
```
npm install
```
Create an uploads directory:
```
mkdir uploads
```
Start the server:
```
node app.js
```

The server will start running on http://localhost:3000.

Docker Setup

Build the Docker image:
```
docker build -t cloudraft-assignment .
```

Run the Docker container:

docker run -p 3000:3000 cloudraft-assignment

The application will be accessible at http://localhost:3000.

Technology Stack

Node.js
Express.js
TensorFlow.js
MobileNet (pre-trained model)
Multer (for file uploads)
Docker (for containerization)

Notes

For simplicity using local storage to to store the downloaded images but in prod should use s3 bucket
Images will be stored within the Docker container's filesystem.
Data will be lost when the container is removed.
To persist data and access it from the host system, we should mount a host directory to the container's uploads directory