Instant Particle Size Distribution Measurement Using CNNs Trained on Synthetic Data

Yasser Eljarida, Youssef Iraqi, Loubna Mekouar · 2025

Read Instant Particle Size Distribution Measurement Using CNNs Trained on Synthetic Data by Yasser Eljarida, Youssef Iraqi, Loubna Mekouar (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/arxiv-2507-00822.txt.

Description

Knowing the spread of particle sizes in a powder matters in mining, drug making, and fertilizer plants. Classic checks, like sieving or laser sizing, are slow, manual, and confused when particles overlap. El Jarida, Iraqi, and Mekouar train neural networks to read sizes straight from images. Their training pictures are not photographs but realistic renders made with the Blender graphics tool, varied in shape, texture, lighting, and layout.

Three network designs were compared on predicting the key size points of a batch. All three scored about equally well. The leanest design, EfficientNet-B0, gave the best speed, suiting live use on a factory line. The authors say well-made synthetic images can train a dependable measuring system without costly photo collection, and they share their code openly.

Catalog record

CREATORYasser Eljarida, Youssef Iraqi, Loubna Mekouar
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE25 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2507-00822

Browse the full library · About the Tower of Babel