Original title: Catalogue Photography as a Cold Start: Toward Deployable Carbide Burr Recognition
Abilash Philip Madavath, Chandra Yuvesh Aubeeluck, Augustin Raju, Nicolas Pyschny +2 · September 2026
Checking that manufactured batches of cutting tools match their order sheets is usually a manual, error-prone task. Factory teams want computer vision to automate this check. However, they face a cold-start problem: they have no labelled photos of the actual tools, only clean manufacturer catalogue pictures. Researchers tested whether an algorithm trained purely on catalogue photos could identify tools in real factory settings. Off-the-shelf image models struggled to distinguish the tools' head shapes and cutting tooth patterns. When the researchers trained the system using metric learning—a method that groups similar items together—it sorted catalogue images almost perfectly. Yet less than half of that success carried over to real field pictures. The biggest fixes were simple adjustments: converting photos to grayscale and matching tools against order sheets using a standard pairing algorithm.
Imagine packing for a trip by looking at clothes under bright store lights. When you wear them outside, the colors look different, making it hard to match outfits. Here, catalogue photos are like store displays—they show tools clearly, but factory conditions alter their appearance. The researchers found that ignoring color and using the order sheet to guide selection helps overcome this mismatch.
Manufacturers spend hours manually checking that tools match order sheets, often making mistakes. This work shows how to start automating that process using only catalogue images, which could reduce errors and save time on factory floors.
The study does not establish a fully deployable system; transfer from catalogue to field remains imperfect. It focuses on specific tool attributes and may not generalize to other features or tools. The evaluation is based on held-out photographs, but real-world deployment could involve additional challenges.
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
Catalogue Photography as a Cold Start: Toward Deployable Carbide Burr Recognition — Abilash Philip Madavath, Chandra Yuvesh Aubeeluck, Augustin Raju et al.
Abstract licence: CC0
Explanation generated by Synthesis from the abstract — it is not the original paper. This is a preprint, so it hasn't necessarily been peer reviewed.