Automating Vehicle Image Sorting for KIA Usedcars with AI-Powered Recognition

  • Customer

    KIA Usedcars

  • Objectives

    Ensure accurate display of vehicle photos on the portal with automated sorting by viewing angles.

  • Challenge

    Manual image uploads by dealers lacked standardized sorting, causing errors in display and confusion for end users.

    To solve this,FMF developed and integrated an AI-powered solution that automatically detects and categorizes vehicle images by35 predefined angles, improving accuracy, efficiency, and user experience.

  • Project Goals
    • Automate image sorting by vehicle angle during uploads via the portal and API.
    • Develop and integrate a deep learning model for recognizing35 vehicle angles with high precision.
    • Achieve at least 95% accuracy in image classification based on extensive training data.
  • Timeline

    65 working days (384 hours)

  • Implementation and Execution

    1

    Research & Analysis

    🔍 Assessed the existing image upload and display process.

    📝 Developed a technical specification outlining AI model requirements and portal integration.

    2

    Database Preparation

    📊 Collected and labeled a training data set of thousands of vehicle images for AI learning.

    3

    AI Model Development

      🧠 Designed a neural network architecture for detecting35 unique vehicle angles.

      📸 Trained the model on over 1,000 images per angle, improving recognition accuracy.

    4

    API Integration & System Deployment

      🔗 Integrated the AI model into KIA Usedcars’ backend system via a REST API.

      ⚡️ Deployed the system on a dedicated server to ensure real-time image processing.

  • Technology Stack

    Programming Language

    Python

    AI Frameworks

    TensorFlow, PyTorch

    API

    REST API integration with KIA Usedcars

    Infrastructure

    Dedicated server for AI model deployment

  • Results

    100%

    Elimination of manual sorting errors
    Dealers no longer need to manually classify images.

    >95%

    AI accuracy in Recognizing
    Vehicle angles and sorting images correctly.

    x15

    Faster image processing

    Before

    Manual sorting took 15 minutes per vehicle

    After

    AI-based sorting completes in less than 1 second per vehicle.

    105,000

    Images processed daily
    Сovering 2000 – 3000 vehicles.

    90%

    Reduction of customer complaints
    For improper display of vehicle images

    By leveraging machine learning and automation, FMF successfully transformed the vehicle image management process for KIA Usedcars. This AI-powered solution significantly improved efficiency, accuracy, and user experience, setting a new standard for digital automotive marketplaces.

Let's discuss your next project