Blog

Retail: semantic catalog search and shelf computer vision

Single-Shot vs Two-Stage vs Transformer Detectors: Why YOLO Wins on the Edge

An object detection model for a retail shelf runs on edge hardware, so single-shot detectors win on speed - two-stage and transformer models are slower.

Search Analytics: The Queries That Tell You What's Wrong With Your Catalog

Ecommerce search analytics is honest customer feedback - zero-results, high-exit, and low-conversion queries each point at a catalog or relevance problem.

Zero-Results Pages: Turning a Dead End Into a Recovery

A zero results search page is a near-certain exit - recover it with labelled relaxed results, a did-you-mean, nearest categories, and by capturing the query.

Store Photos and Privacy: Handling Staff and Customers in Shelf Images

Retail computer vision privacy: shelf photos catch people incidentally. Process on-device, keep only the result, retain no images, add no face recognition.

What Does It Cost to Deploy Shelf-Monitoring Computer Vision Per Store?

Computer vision cost per store: a one-time build, a per-store rollout for hardware and install, and ongoing retraining, fleet management, and review staffing.

How to Measure ROI on Computer Vision for Shelf Availability

Computer vision retail ROI: recovered sales from faster gap closure, audit hours saved, and a store holdout to measure it net of substitution.

Common Mistakes When Rolling Out Store Computer Vision Across a Chain

Retail computer vision rollout mistakes: designing for connectivity stores lack, a single confidence threshold, stale training data, and a flagship-only pilot.

How a Shelf-Monitoring Computer Vision Pipeline Is Architected

Shelf-monitoring computer vision architecture: on-device detection, a portable runtime, pub-sub transport, and confidence-based routing to a review queue.

Edge vs Cloud Inference for In-Store Computer Vision

Edge vs cloud computer vision for retail: store links can't upload shelf photos fast enough, so detection runs on-device and only the result leaves the store.

Giving Merchandising Control of Search Relevance Without Engineering Tickets

Merchandising control of search relevance: boost, bury, pin and filter rules stored as data the team edits directly, with owners, expiries, and an audit log.