Smile Detection

FastAPI-based Image Processing Application

Overview

Smile Detection is a FastAPI-based web application that allows users to upload images for processing using AI models. The application supports file uploads, processes images, and returns results to the client.

Features

• Upload PNG and JPG images for processing
• AI-based image analysis using a pre-trained model
• Serve processed files for download
• Simple HTML-based frontend using Jinja2 templates
• Static file handling via FastAPI

Installation

Prerequisites

• Python 3.8+
• FastAPI
• Uvicorn (for running the server)
• OpenCV (cv2) for image processing
• NumPy for numerical computations
• Joblib for loading the AI model

Setup Instructions

1. Clone the repository:

   git clone https://github.com/mhdned/SmileDetection.git
   cd SmileDetection

2. Create a virtual environment:

   python -m venv venv
   source venv/bin/activate  # On Windows use `venv\Scripts\activate`

3. Install dependencies:

   pip install -r requirements.txt

4. Run the application:

   uvicorn main:app --reload

API Endpoints

Check API Status

Endpoint: GET /

• Returns an HTML page confirming that the API is live.

Upload and Process Image

Endpoint: POST /process

• Accepts an image file (PNG, JPG) and processes it using the AI model.
• Returns an HTML response with the processed result.

Download Processed File

Endpoint: GET /file/{file_name}

• Allows users to download processed files.

AI Model

The application utilizes an AI model stored in smile.z, which is loaded using Joblib. The model is used to analyze images and classify them accordingly. The AI logic is implemented in core/ai.py:

import cv2
from joblib import load
import numpy as np

clf = load("smile.z")

def load_pic(item):
    img = cv2.imread(item)
    img_r = cv2.resize(img,(32,32))
    img_r = img_r / 255
    img_r = img_r.flatten()
    img_r = np.array([img_r])

    label = clf.predict(img_r)[0]
    return label

Project Structure

├── main.py            # Main application file
├── core/
│   ├── ai.py          # AI model integration
│   ├── jinja.py       # Jinja template engine configuration
├── static/            # Static files (CSS, JS, images)
├── templates/         # HTML templates
├── uploads/           # Uploaded images
├── requirements.txt   # Python dependencies
├── README.md          # Project documentation

Technologies Used

License

This project is licensed under the MIT License. See LICENSE for details.

Author

Developed by Mehtiuo & Sudotome.