Mean shift based prototypical network for steel surface anomaly recognition
Canan Tastimur · 2025
Read Mean shift based prototypical network for steel surface anomaly recognition by Canan Tastimur (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-12749279.txt.
Description
Canan Tastimur presents a defect finder for steel sheets in a 2025 paper. Hot-rolled steel with surface flaws loses value and buyer trust. Factories want flaws caught early in the making process. The new method, MSPro-Net, sorts flaw photos into classes. It builds on a network style that learns a model example, called a prototype, for each class.
The twist is how the prototype is picked. A mean shift method, which nudges the center toward the densest spot, gives a prototype that fits all class members better. The method shines when training photos are few. On one set with six classes and 25 shots each, it hit 98.67 percent accuracy. The older style reached only 50.67 percent on the same test. Two more sets showed like gains.
Catalog record
| CREATOR | Canan Tastimur |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 64 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |