shelf audit

Refining product ID: When similar packages confuse embeddings, what should Stage 2 do?

A shelf audit tool lives or dies on telling a 1.25L bottle from a 2L one, and that's exactly where this project hits a wall. Cropping works, but the embedding stage fails because letterboxing scrubs away the tiny size…

4 min readMachine Learning
From Machine Learning

Help: Project l'm building a shelf audit tool. A photo goes through YOLO, which crops each product, and then I embed the crop and search a small gallery of reference photos to get the SKU. New products should be addable by just dropping in photos, no detector retrain. Detection is basically fine. ldentification is not. Same brand, same bottle, different flavor or size (like 1.25 L vs 2 L) and the nearest neighbor is often the wrong SKU. Correct and wrong scores overlap, so a threshold either misses real products or accepts the wrong one. I tried DINOV2, SigLIP2 and…

Read the original at Machine Learning