Improves product discovery and customer engagement
Retail & Consumer
|
Product
|
Computer Vision
Learning efficient visual representations to improve product image search and recommendations.
Etsy’s AI looks at product photos and learns to group similar-looking items so shoppers can find what they like faster—even if they don’t know what it’s called.
Etsy developed a visual representation learning system to better understand and process product images for search and recommendation features. By training a model to extract high-quality, compact visual embeddings, Etsy can more efficiently compare products visually, even across a massive and diverse catalog. The approach boosts the accuracy and relevance of visual similarity searches while keeping computational costs manageable, making it easier for shoppers to find visually similar items.
It’s like teaching your phone to recognize the difference between vintage boots and modern sneakers just by looking—then helping you shop for more of what you like.
3
/5
Visual search and recommendations are now common in e-commerce, with moderate process innovation.
Timeline:
10 months
Cost:
$1,350,000
Headcount:
8