Increasing Revenue and Enhancing Customer Experience with Computer Vision

Challenge  
The client faced the challenge of making their online shop more attractive and increasing the conversion rate. Product images were often poorly described, which negatively impacted search and filtering functionalities. Additionally, inaccurate product recommendations hindered the customer experience. Without an efficient solution, there was a risk of losing customers to competitors and missing out on potential revenue.  

Approach
We developed a computer vision suite based on deep learning models, tailored specifically for the client during a 4-week project.

The solution included:  

• Image Classification: Automatic categorization of product images by type, brand, and scenarios.  
• Object Detection: Identification of specific product features to enrich metadata.  
• Targeted Product Recommendations: Analysis of preferences among specific customer groups to identify the most appealing product images and items.  

The suite was deployed on a scalable AWS infrastructure to ensure reliable and cost-efficient live usage.  

Results  
The results of the project were remarkable:  

• Increased Purchase Probability: By analyzing which images resonate best with target audiences, the conversion rate significantly improved.  
• Enhanced Metadata: Automated labeling for new and existing product images improved search and filtering functionalities in the online shop.  
• Scalable Solution: The developed computer vision suite operates seamlessly in production, enabling the client to continuously benefit from its insights.