A large US-based retailer wanted to digitize all its 20,000 retail stores on specific categories, including over 1200 SKUs/ products across nine categories.
Solution:
We used SmartProject© and SmartAIX© to support the entire project. Divided into multiple phases, this project was implemented over ten months. The project began with AI training for six categories, and over time, we added more categories, implemented label and price detection barcode detection, and focused on reducing invalidity. We iteratively trained the AI to take care of edge cases. We used SmartProject© to manage the entire execution, while SmartAIX© was used for the AI functionality.
Outcome:
We achieved 95% accuracy for SKU recognition within 30 days and moved the accuracy to 97% towards the end of the project. The solution can correct image tilt up to 25 degrees, detect blur, and has built capacity to process 50,000 images per day.
