Department of Mechanical, Materials and Manufacturing Engineering, University of Nottingham, University Park, Nottinghamshire, UK.
J Microsc. 2022 Oct;288(1):28-39. doi: 10.1111/jmi.13140. Epub 2022 Sep 16.
A pre-trained convolution neural network based on residual error functions (ResNet) was applied to the classification of soot and non-soot carbon nanoparticles in TEM images. Two depths of ResNet, one 18 layers deep and the other 50 layers deep, were trained using training-validation sets of increasing size (containing 100, 400 and 1400 images) and were assessed using an independent test set of 200 images. Network training was optimised in terms of mini-batch size, learning rate and training length. In all tests, ResNet18 and ResNet50 had statistically similar performances, though ResNet18 required only 25-35% of the training time of ResNet50. Training using the 100-, 400- and 1400-image training-validation sets led to classification accuracies of 84%, 88% and 95%, respectively. ResNet18 and ResNet50 were also compared for their ability to categorise soot and non-soot nanoparticles via a fivefold cross-validation experiment using the entire set of 800 images of soot and 800 images of non-soot. Cross-validation was repeated 3 times with different training durations. For all cross-validation experiments, classification accuracy exceeded 91%, with no statistical differences between any of the network trainings. The most efficient network was ResNet18 trained for 5 epochs, which reached 91.2% classification after only 84 s of training on 1600 images. Use of ResNet for classification of 1000 images, the amount suggested for reliable characterisation of soot sample, requires <4 s, compared with >30 min for a skilled operator classifying images manually. Use of convolution neural networks for classification of soot and non-soot nanoparticles in TEM images is highly promising, particularly when manually classified data sets have already been established.
基于残差函数（ResNet）的预训练卷积神经网络被应用于 TEM 图像中 soot 和非 soot 碳纳米颗粒的分类。使用包含 100、400 和 1400 张图像的训练-验证集训练了两个深度的 ResNet，一个 18 层深，另一个 50 层深，并使用包含 200 张图像的独立测试集进行评估。网络训练在小批量大小、学习率和训练长度方面进行了优化。在所有测试中，ResNet18 和 ResNet50 的性能统计上相似，尽管 ResNet18 只需要 ResNet50 训练时间的 25-35%。使用 100、400 和 1400 张图像的训练-验证集进行训练，分类准确率分别为 84%、88%和 95%。还通过使用整个 800 张 soot 图像和 800 张 non-soot 图像的 5 倍交叉验证实验，比较了 ResNet18 和 ResNet50 区分 soot 和 non-soot 纳米颗粒的能力。交叉验证重复了 3 次，每次使用不同的训练持续时间。对于所有交叉验证实验，分类准确率均超过 91%，并且在任何网络训练中均无统计学差异。最有效的网络是经过 5 个时期训练的 ResNet18，在对 1600 张图像进行 84 秒的训练后，达到了 91.2%的分类准确率。对于可靠地表征 soot 样本所需的 1000 张图像的分类，使用 ResNet 只需要 <4 秒，而对于熟练操作员手动分类图像，则需要 >30 分钟。使用卷积神经网络对 TEM 图像中的 soot 和 non-soot 纳米颗粒进行分类具有很高的前景，特别是在已经建立了手动分类数据集的情况下。
