Region versus query based instance segmentation models: application to the estimation of aggregated TiO2 particles size distribution measured by SEM

Instance segmentation model performances is commonly evaluated through challenges using toy datasets such as CIFAR and ImageNet. Although, it is necessary to help improving state-of-the-art models, it remains far from critical industrial application where accurate measurement is required to ensure o...

Full description

Saved in:
Bibliographic Details
Main Authors: Paul Monchot, Loïc Coquelin, Nicolas Fischer
Format: Article
Language:English
Published: IOP Publishing 2025-01-01
Series:Machine Learning: Science and Technology
Subjects:
Online Access:https://doi.org/10.1088/2632-2153/add239
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849328685070417920
author Paul Monchot
Loïc Coquelin
Nicolas Fischer
author_facet Paul Monchot
Loïc Coquelin
Nicolas Fischer
author_sort Paul Monchot
collection DOAJ
description Instance segmentation model performances is commonly evaluated through challenges using toy datasets such as CIFAR and ImageNet. Although, it is necessary to help improving state-of-the-art models, it remains far from critical industrial application where accurate measurement is required to ensure optimal process control, quality assurance and safety. This paper compares the performance of ready-to-deploy state-of-the-art instance segmentation models, specifically region and query-based models, across COCO and task-specific designed metrics in the context of the estimation of aggregated TiO _2 particle size distribution from scanning electron microscopy (SEM) measurements. The numerical comparison is conducted on custom datasets, addressing well-known challenges of SEM microscopists: high resolution images showing a massive number of overlapping instances, images measured with diverse imaging parameters (operator control) or the presence of a mixture of aggregated particles. Our findings highlight the strengths and limitations of each model category, providing insights into their suitability for high-stakes industrial deployment. Finally, we propose a customized inference procedure that leverages the performance of transformers on images that contain a small number of instances within dense and clustered scenes.
format Article
id doaj-art-8b060ff0628147369f3e5fbdc1b9e6bc
institution Kabale University
issn 2632-2153
language English
publishDate 2025-01-01
publisher IOP Publishing
record_format Article
series Machine Learning: Science and Technology
spelling doaj-art-8b060ff0628147369f3e5fbdc1b9e6bc2025-08-20T03:47:32ZengIOP PublishingMachine Learning: Science and Technology2632-21532025-01-016202050210.1088/2632-2153/add239Region versus query based instance segmentation models: application to the estimation of aggregated TiO2 particles size distribution measured by SEMPaul Monchot0https://orcid.org/0000-0002-6658-1798Loïc Coquelin1https://orcid.org/0000-0001-5834-252XNicolas Fischer2https://orcid.org/0000-0003-3628-8117Data Science and Uncertainty Department, National Laboratory of Metrology and Testing , 29 avenue Roger Hennequin, 78197 Trappes, FranceData Science and Uncertainty Department, National Laboratory of Metrology and Testing , 29 avenue Roger Hennequin, 78197 Trappes, FranceData Science and Uncertainty Department, National Laboratory of Metrology and Testing , 29 avenue Roger Hennequin, 78197 Trappes, FranceInstance segmentation model performances is commonly evaluated through challenges using toy datasets such as CIFAR and ImageNet. Although, it is necessary to help improving state-of-the-art models, it remains far from critical industrial application where accurate measurement is required to ensure optimal process control, quality assurance and safety. This paper compares the performance of ready-to-deploy state-of-the-art instance segmentation models, specifically region and query-based models, across COCO and task-specific designed metrics in the context of the estimation of aggregated TiO _2 particle size distribution from scanning electron microscopy (SEM) measurements. The numerical comparison is conducted on custom datasets, addressing well-known challenges of SEM microscopists: high resolution images showing a massive number of overlapping instances, images measured with diverse imaging parameters (operator control) or the presence of a mixture of aggregated particles. Our findings highlight the strengths and limitations of each model category, providing insights into their suitability for high-stakes industrial deployment. Finally, we propose a customized inference procedure that leverages the performance of transformers on images that contain a small number of instances within dense and clustered scenes.https://doi.org/10.1088/2632-2153/add239scanning electron microscopybenchmarktransformersdeep learningparticle size distributioninstance segmentation
spellingShingle Paul Monchot
Loïc Coquelin
Nicolas Fischer
Region versus query based instance segmentation models: application to the estimation of aggregated TiO2 particles size distribution measured by SEM
Machine Learning: Science and Technology
scanning electron microscopy
benchmark
transformers
deep learning
particle size distribution
instance segmentation
title Region versus query based instance segmentation models: application to the estimation of aggregated TiO2 particles size distribution measured by SEM
title_full Region versus query based instance segmentation models: application to the estimation of aggregated TiO2 particles size distribution measured by SEM
title_fullStr Region versus query based instance segmentation models: application to the estimation of aggregated TiO2 particles size distribution measured by SEM
title_full_unstemmed Region versus query based instance segmentation models: application to the estimation of aggregated TiO2 particles size distribution measured by SEM
title_short Region versus query based instance segmentation models: application to the estimation of aggregated TiO2 particles size distribution measured by SEM
title_sort region versus query based instance segmentation models application to the estimation of aggregated tio2 particles size distribution measured by sem
topic scanning electron microscopy
benchmark
transformers
deep learning
particle size distribution
instance segmentation
url https://doi.org/10.1088/2632-2153/add239
work_keys_str_mv AT paulmonchot regionversusquerybasedinstancesegmentationmodelsapplicationtotheestimationofaggregatedtio2particlessizedistributionmeasuredbysem
AT loiccoquelin regionversusquerybasedinstancesegmentationmodelsapplicationtotheestimationofaggregatedtio2particlessizedistributionmeasuredbysem
AT nicolasfischer regionversusquerybasedinstancesegmentationmodelsapplicationtotheestimationofaggregatedtio2particlessizedistributionmeasuredbysem