AWS Only
Streamlining Vehicle Image Classification with Generative AI at Groupe La Centrale
Groupe La Centrale faced a challenge with manually auditing vehicle photos for ad listings, a time-consuming process governed by basic criteria. To address this, Firemind developed a generative AI-powered proof of concept that automates the classification and labelling of vehicle images. This solution efficiently identifies key attributes such as make, model, colour, and orientation, significantly speeding up the review process.
AWS ONLY ·
Streamlining Vehicle Image Classification with Generative AI at Groupe La Centrale
Groupe La Centrale faced a challenge with manually auditing vehicle photos for ad listings, a time-consuming process governed by basic criteria. To address this, Firemind developed a generative AI-powered proof of concept that automates the classification and labelling of vehicle images. This solution efficiently identifies key attributes such as make, model, colour, and orientation, significantly speeding up the review process.
At a glance
Groupe La Centrale is a French platform for buying and selling automobiles and motorcycles, featuring native vehicle listings and ad ingestion from partner platforms, facilitating transactions across a wide range of vehicles.
Challenge
Groupe La Centrale struggled with the manual and time-consuming process of auditing vehicle photos for ad listings, which involved basic criteria and was inefficient for managing large volumes of images.
Solution
Firemind created a generative AI-powered tool to automate vehicle photo classification and labelling, efficiently handling attributes like make, model, colour, and orientation, and streamlining the photo review process for Groupe La Centrale.
Services used
Amazon Bedrock
Amazon S3
Amazon DynamoDB
AWS Step Functions
Outcomes
Improved accuracy and consistency in vehicle photo classification and labelling.
Significantly reduced time and manual effort required for photo auditing.
Business challenges
Inefficient and time-consuming manual vehicle photo auditing
Groupe La Centrale, a leading French platform for buying and selling vehicles, faced a significant challenge with their manual photo auditing process. Each vehicle listing required detailed scrutiny of images to ensure accuracy in attributes such as make, model, colour, and orientation. This manual process was not only time-consuming but also prone to human error, leading to inconsistent and inefficient handling of large volumes of vehicle photos.
The current system relied on basic criteria for photo assessment, which was insufficient for managing the complexities of vehicle listings. As a result, the platform struggled with delays and inefficiencies, impacting overall operational performance and user experience. To address these issues, a solution was needed that could automate and streamline the photo review process while maintaining high accuracy and consistency.
Enter
Add role
Solution
Automated generative AI solution for efficient vehicle photo classification
To address Groupe La Centrale’s challenge of manual and inefficient vehicle photo auditing, we developed a comprehensive generative AI-powered solution leveraging several key AWS services. Our solution automates the classification and labelling of vehicle images, significantly enhancing both speed and accuracy in the review process.
We utilised Amazon S3 to securely store and manage the vast array of vehicle images provided by Groupe La Centrale. The images were ingested from S3 into our analysis pipeline, where they were processed and classified. To perform the image analysis, we employed AWS Bedrock, which enabled us to leverage advanced generative AI capabilities for identifying and labelling critical attributes such as make, model, colour, and orientation of each vehicle.
The processed data, including image metadata and labels, was then stored in Amazon DynamoDB. This NoSQL database provided a scalable and efficient way to organise and retrieve the classified information based on job and vehicle IDs. Finally, AWS Step Functions orchestrated the workflow, ensuring seamless integration between the various stages of the image processing pipeline and facilitating the smooth execution of tasks from ingestion to classification.
Together, these AWS services enabled us to deliver a solution that streamlined the photo review process, reduced manual effort, and improved the overall accuracy of vehicle image classification.
Enhanced accuracy
The generative AI solution consistently met or exceeded the accuracy of previous ML models, ensuring reliable classification and labelling of vehicle images.
Increased efficiency
The automated process significantly reduced the time required for photo auditing, streamlining operations and enabling quicker turnaround for vehicle listings.
Improved consistency
Automation provided consistent results across all vehicle images, eliminating human error and standardising the review process.
Model Spotlight
Claude 3 Sonnet V1
For this project, we utilised the Claude 3 Sonnet model to analyse images with high precision, focusing primarily on visual data rather than text. The model was customised to handle specific queries, providing structured answers tailored to the needs of the task.
It demonstrated the ability to accurately identify and count cars within images, assess any visible damage, and determine the make, model, and colour of the vehicles. Additionally, it could distinguish whether the image showed the front or rear of the car and classify whether the photo was taken indoors or outdoors. This tailored image analysis capability made Claude 3 Sonnet an ideal solution for our project requirements.
Why Firemind
Firemind was chosen as a partner for this project due to its expertise in leveraging advanced technologies like generative AI and its experience with AWS services. Firemind’s capability to build and deploy scalable solutions using AWS tools such as Amazon S3, Amazon Bedrock, Amazon DynamoDB, and AWS Step Functions aligns well with Groupe La Centrale’s needs for automating and streamlining the vehicle photo auditing process. Additionally, Firemind’s track record of delivering high-quality, efficient solutions that address complex challenges in data processing and machine learning likely contributed to their selection.
100%
Reduction in Manual Data Processing
The automated solution eliminated the need for manual data scraping and analysis, significantly reducing time and effort.
50%
Scalability increase
The AI-driven system enabled Blue Light Card to scale their competitive analysis operations by 50% without adding additional full-time employees (FTEs).
Added value
The AI-driven system provided Blue Light Card with deeper insights and more accurate data, enabling them to make informed, data-driven decisions. With the ability to rapidly process and analyse competitor information, Blue Light Card could quickly adapt to market trends and stay ahead of the competition.
Get in touch
Want to learn more?
Seen a specific case study or insight and want to learn more? Or thinking about your next project? Drop us a message below!