Performance comparison among VGG16, InceptionV3, and resnet on galaxy morphology classification
Yumeng Qian
Abstract
Abstract This article introduces the structures of three classical convolutional neural networks: VGG16, InceptionV3, and ResNet50, and compares their performance on galaxy morphology classification. The different structures of these networks give them some distinct features, leading to different results. The dataset used was Galaxy10 DECals created by the Galaxy Zoo project. The models were compared in the same way, using one fully-connected layer only without modifying its original architecture, evaluated by accuracy, precision, recall rate, and f1 score. The experimental results show that InceptionV3 gave the best classification results from all aspects. The inception module performs well on the galaxy morphology classification problem.