Nordic Microscopy Society, SCANDEM2023, will be organized at Uppsala, Sweden, 12-15 June 2023.
We present materials characterization using deep learning image analysis.
We present materials characterization using deep learning image analysis.
The quick development of materials relies on the understanding of material microstructure
(size, porosity, morphology, etc.), a reliable manufacturing process, and a thorough analysis of
the performance for different applications. To solve some of the research challenges, automated,
reliable, and intelligent analysis techniques are needed.
Using deep learning and a powerful image analysis engine, MIPAR [1] (www.mipar.us) allows
users to perform fast, accurate, and automated analysis of images. In three simple steps: trace,
train and apply, researchers can create a model that identifies the features of interest and run
personalized recipes on new images to detect complex features.
This presentation will overview the advantages of using modern analysis techniques to analyze
particles, fibers and pores, droplets, defects, contaminants, and phases with real research
applications.
Figures/Tables
Figure 1 Detecting grains while ignoring twins using MIPAR. Model was trained on 25 images
in 40 minutes and applied to new images in 2 seconds
Figure 2 Overlapping nanofiber network analysis performed using MIPAR. Model was trained
on 36 images in 40 minutes and applied to new images in 1.5 seconds.
References
[1] Sosa, J. M., et al. “MIPAR™: 2D and 3D Image Analysis Software Designed by
Materials Scientists, for All Scientists.” Microscopy and Microanalysis 23.S1 (2017):
230-231.
