During the COVID-19 pandemic, recommendations for wearing medical masks have been adopted around the world and admission to crowded places of those who ignore this requirement is limited. With the help of convolutional neural networks and deep learning, it is possible to analyze the video stream to facilitate work at checkpoints and mark people without masks. This article examines three neural networks, their training and comparative analysis. This example shows how you can easily prepare a neural network for a specific task.
neural network, deep learning, CNN, machine learning, computer vision, MobileNet V2, Xception, Inception V3
1. Freymvork Tensorflow2. URL:https://www.tensorflow.org/ (data obrascheniya 28.07.2021).
2. Osnovnye svedeniya o COVID-19URL: https://www.who.int/ru/emergencies/diseases/novel-coronavinis-2019/question-and-an-swers-hub/q-a-detail/coronavimse-disease-covid-19(data obrascheniya 28.07.2021).
3. Glubokoe obuchenie: razbiraemsya so svertkami. https://medium.com/@balovbohdan/%D0%B3%D0%BB%D1%83%D0%B1%D0%BE%D0%BA%D0%BE%D0%B5-%D0%BE%D0%B1%D1%83%D1%87%D0%B5%D0%BD%D0%B8%D0%B5-%D1%80%D0%B0%D0%B7%D0%B1%D0%B8%D1%80%D0%B0%D0%B5%D0%BC%D1%81%D1%8F-%D1%81%D0%BE-%D1%81%D0%B2%D0%B5%D1%80%D1%82%D0%BA%D0%B0%D0%BC%D0%B8-6e47bfc27792 (data obrascheniya 29.07.2021).
4. Vvedenie v svertochnye neyronnye seti (S onvolutionalNeuralNet works). URL: https: //habr.com/ni/post/454986/(data obrascheniya 29.07.2021).
5. Face Mask Detection ~12K Images Dataset. URL: https://www.kaggle.com/ashishiangra27/face-mask-12k-images-datasetf data obrascheniya 29.07.2021).
6. MobileNet: men'she, bystree, tochnee. URL: https://habr.com/m/post/352804/(flaTa obrascheniya 29.07.2021).
7. MobileNetV2: invertirovannye ostatki i lineynye uzkie mesta. URL: https://www.nia-chineleamingmasterv.nl/mobilenetv2-inverted-re-siduals-and-linear-bottlenecks-8a4362f4ffd5/( data obrascheniya 30.07.2021).
8. Xception: kompaktnaya glubokaya neyronnaya set'. URLhttps://habr.com/m/post/347564/(flaTa obrascheniya 30.07.2021).
9. ObzorHception. ULL: httpsV/www.niachineleani-ingmasterv.ru/review-xception-with-depthwise-separable-convolution-better-than-inception-v3-image-dc967dd42568/('flaTa obrascheniya 30.07.2021).
10. A Simple Guide to the Versions of the Inception Network. URL: https://towardsdatascience.com/a-simple-guide-to-the-versions-of-the-inception-net-work-7fc52b863202f data obrascheniya 31.07.2021).
