This repository is part of a larger project aimed at evaluating damage on cars. Currently, this repository includes the following components:

The image scraping component retrieves images from three websites: AutoMotos, SalvageReseller, and Schadeautos. The scrapers are located in the ImageCarsScrapping directory.

• AutoMotos: A website with a variety of car images.
• SalvageReseller: A site specializing in images of cars that are for sale as salvage.
• Schadeautos: A site with images of damaged cars.

• Gemini API Annotation Script: This script performs image annotation using the Gemini API and OpenAI. It fetches images from an S3 bucket, processes them using a generative model, and saves the results to a CSV file.

• OpenAI API Annotation Script: This script performs image annotation using the OpenAI API. It fetches images from an S3 bucket, processes them using a GPT-4 model to identify car parts and their conditions, and saves the results to a CSV file.

The cleaning application, located in the cleaningApp directory, is built with Flask (for the backend) and JavaScript, HTML, and CSS (for the frontend). This application allows users to annotate images.

• Predefined Annotations: The application uses predefined annotations generated based on Gemini & GPT.
• User Validation: Users can validate or correct the annotations provided by LLMs.

• ImageCarsScrapping/: Contains scripts for scraping images from the specified websites.
• cleaningApp/: Contains the Flask application for annotating images.

After completing the annotation of damage regions in the car images, the next steps for the project include:

• Design and implement a machine learning model to predict the location and extent of damage on cars based on the annotated data.

• Create a segmentation model to accurately delineate damaged areas on the car images, which will help in providing more detailed damage assessments.

• Python 3.x
• Flask
• Datrie
• boto3
• JavaScript, HTML, and CSS for the frontend
• Necessary Python libraries listed in requirements.txt

1. Clone the repository:

   git clone [email protected]:RChoukri03/CarDamageDetection-.git
   cd CarDamageDetection