Model AI Car Recognition & Price Suggestion
**Issue:** The customer's website recommends prices for used cars based on user-submitted information, but the extensive data source makes it time-consuming for users to find and select options. This inefficient online experience results in higher bounce rates.
**HBLAB's Solution:** Enhance the user experience by implementing a solution that automatically imports data using just a photo. Users simply need to upload images, and the majority of the required information will be automatically analyzed and populated in the system. An AI model accurately fills in details for 250 car types, including color, model, and year of manufacture. This AI system also analyzes and suggests car prices based on experience with over 300,000 similar market products. It is deployed on AWS Cloud, offering flexibility in scaling and feature adaptation.
**Project Details:**
- **Development of the Recognition AI Model:**
- Team: 1 AI engineer, 1 Data Crawler
- Duration: 2 months
- **Development of the Price Suggestion AI Model:**
- Team: 2 AI engineers, 1 Data Crawler
- Duration: 3 months
- Technologies Used: Object Detection (YOLOV4, ResNet50) and Deep Learning
**Outcome:** The AI model automatically inputs information for 250 car types with 95% accuracy, categorizing attributes like color, model, and year of manufacture. The suggested car value has a maximum price discrepancy of 20,000 Yen (~$160), depending on the accuracy of the information provided and the vehicle's price classification.