Pre-trained Model for Image Prediction Using Keras Application

Pre-trained model is used in transfer learning to solve the new problem. Pre-trained model generally used for image prediction, classification, text generation, fill mask, machine translation etc., as the data is more machine learning performance may become stagnant to resolve this challenge deep learning came into existence. Deep learning performance is relatively higher than machine learning model. Deep learning makes use of the artificial neural network which enables the model to learn the features by its own with the minimal intervention of the user. This paper focus more on EfficientNetB7, InceptionV3, MobileNet, MobileNetV2, Xception, and ResNet50 trained model for image prediction. MobileNet, MobileNetV2, EfficientNetB7, Xception, InceptionV3and ResNet50 model is implemented using keras application. The result indicates that EfficientNetB7 performance is better over the other trained model.