A Deep Learning-Based Cluster Analysis Method for Large-Scale Multi-Label Images

Abstract

Large-scale multi-label image classification requires determining the presence or absence of a target object in a large number of sample images. For highly specialized and complex multi-label image sets, it is especially important to ensure the accuracy of image classification. Traditional deep learning models usually don’t take into account image-label correlation constraints when classifying multi-label images, and the strategy of classifying images based only on their own features greatly limits the model performance. In this context, this paper focuses a deep learning-based cluster analysis method for large-scale multi-label images. We constructed a model for large-scale multi-label image category recognition, which consists of a global image feature extraction module, a feature activation vector generation module and an image category inter-label connection module. Using a graph convolutional network (GCN), we aggregated the information of image category label nodes in the constructed multi-label graph structure, while exploring the correlation between image category labels. A detailed description is presented on how to introduce the attention mechanism into the constructed model mentioned above for image category recognition. Experimental results have validated the effectiveness of the constructed model.

A Deep Learning- Image Based Approach for Detecting Cracks in Buildings

“…Consequently, if the results are observed carefully, RNN attempts to overfit the entire dataset and fails miserably by achieving test accuracy of only 50 percent, whereas CNN achieves test and train accuracies that are equivalent. For the dataset, it can be inferred that CNN has an accuracy of 400 misclassified images out of 40,000, of which 350 are from the training side and 50 are from the testing side [39][40][41][42]. Table 1 depicts the performance of the proposed work in training, testing, and prediction.…”