How GenAI is transforming e-commerce catalogues
summary
Target describes a deployed stack that mixes classical ML, vision models and GenAI to clean and enrich millions of SKUs: anomaly detection for package dimensions using a XGBoost classifier, image decisioning that segments products with SAM and reads overlays with PaddleOcr, attribute recommendation and extraction via GenAI, training on Vertex AI with PyTorch, API/Kafka integrations and an accuracy framework that scales derived attributes to 95 categories, cutting vendor onboarding and time‑to‑production substantially.
AI-written summary · about
chapters (10)
- 0:00GenAI in e-commerce catalogues
- 3:09Tech for item data challenges
- 5:37Suggestive variation for grouping
- 7:03AI for image policy violations
- 9:08Generating infographics with GenAI
- 12:01Derived attribution for product data
- 17:42Accuracy framework for AI enrichment
- 21:39Accurate item data impact
- 22:42Key takeaways and tech choice
- 28:19Q&A on attribute enrichment
related talks
16mGen AI For Retail Data Management
26mInnovate faster with generative AI
37mLeveraging Generative AI for Enhanced E-commerce
09mGen AI for regulated industries
29mCustomizing genAI applications for your business
23mScaling GenAI Applications and Agentic AI for Business Competitive Differentiation
30mAI in Retail - Advancing Promo Demand Forecasting
42mThe Rise of AI