Application of Transfer Learning for Image Classification on Dataset with Not Mutually Exclusive Classes

Machine learning technologies, especially deep convolution neural network (CNN), play an important role in image classification tasks. However, performing image classification tasks using state-of-the-art deep learning models might suffer from the lack of available images for network training and requirement of computationally powerful machines to conduct the training. In order to classify new classes, in this paper, transfer learning models are built based on the pretrained AlexNet and the VGG16 to overcome the drawbacks of the deep CNN. The models are used on a not well-classified image dataset, where classes of the images are not mutually exclusive, and an image could belong to more than one classes. Experimental results are given to evaluate the performance of the transfer learning approach on this not exclusive dataset, and the conventional CNN are used as the benchmark. It shows that the transfer learning models outperform the conventional CNN by a large margin in both coupled and decoupled datasets.