Convolutional Neural Networks with Transfer Learning for Recognition of COVID-19: A Comparative Study of Different Approaches

摘要: To judge the ability of convolutional neural networks (CNNs) to effectively and efficiently transfer image representations learned on ImageNet dataset task recognizing COVID-19 in this work, we propose analyze four approaches. For purpose, use VGG16, ResNetV2, InceptionResNetV2, DenseNet121, MobileNetV2 CNN models pre-trained extract features from X-ray images COVID Non-COVID patients. Simulations study performed by us reveal that these have a different level representation. We find approaches proposed, if either ResNetV2 or DenseNet121 features, then performance detect is better. One important findings our principal component analysis for feature selection improves efficiency. The approach using fusion outperforms all other approaches, with approach, could achieve an accuracy 0.94 three-class classification problem. This work will not only be useful detection but also any domain small datasets.