Convolution neural network for classifying TEM images of soot
An ONNX file - associated with Sipkens et al. (https://doi.org/10.1016/j.powtec.2021.04.026) - containing a trained convolution neural network to separate soot aggregates from the background in TEM images. The network was trained on images collected at the University of British Columbia for soot collected from a laboratory-scale flare, a marine engine, and an internal combustion engine. Please see the associated manuscript for more information on architecture and validation.
Steps to reproduce
See the repository at https://github.com/maxfrei750/CarbonBlackSegmentation for the open-source code used to train this network.