Training Convolutional Neural Networks with Multi-Size Images and Triplet Loss for Remote Sensing Scene Classification

Abstract

Many remote sensing scene classification algorithms improve their classification accuracy by additional modules, which increases the parameters and computing overhead of the model at the inference stage. In this paper, we explore how to improve the classification accuracy of the model without adding modules at the inference stage. First, we propose a network training strategy of training with multi-size images. Then, we introduce more supervision information by triplet loss and design a branch for the triplet loss. In addition, dropout is introduced between the feature extractor and the classifier to avoid over-fitting. These modules only work at the training stage and will not bring about the increase in model parameters at the inference stage. We use Resnet18 as the baseline and add the three modules to the baseline. We perform experiments on three datasets: AID, NWPU-RESISC45, and OPTIMAL. Experimental results show that our model combined with the three modules is more competitive than many existing classification algorithms. In addition, ablation experiments on OPTIMAL show that dropout, triplet loss, and training with multi-size images improve the overall accuracy of the model on the test set by 0.53%, 0.38%, and 0.7%, respectively. The combination of the three modules improves the overall accuracy of the model by 1.61%. It can be seen that the three modules can improve the classification accuracy of the model without increasing model parameters at the inference stage, and training with multi-size images brings a greater gain in accuracy than the other two modules, but the combination of the three modules will be better.

Diverse Capsules Network Combining Multiconvolutional Layers for Remote Sensing Image Scene Classification

“…However, Although CNN-based high-level features have achieved the state-of-the-art results [25]- [41], they usually lose more details that are crucial to complex image scene classification after the pooling operations in CNN. Therefore, how to extract the discriminative and robust features containing detailed information and salient objects in the scenes is a key issue for scene classification.…”

“…Yuan et al [40] proposed pyramid multi-subset weighted multi-deep feature fusion (PMWMFF) to effectively fuse the deep features extracted from different pretrained CNNs and integrate the global and local information of the deep features. Zhang et al [41] proposed a network training strategy with multi-size images by triplet loss. Cheng et al [35] propose a simple but effective method to learn discriminative CNNs (D-CNNs) to boost the performance of remote sensing image scene classification.…”

The image inpainting algorithm used on multi-scale generative adversarial networks and neighbourhood

“…With the further development of deep learning [58][59][60] [37][38][39], the Generative Adversarial Networks (GAN) have been proposed by the literature [40]. The literature [40] is a milestone in the development of deep learning.…”

Non Invasive Skin Hydration Level Detection Using Machine Learning

“…However, the BIA is a complex method which cannot be used for monitoring and it is only limited for an indirect measure of hydration level (HL). In the field of computer vision and many other applications deep neural network provides impressive result [17][18][19]. Convolutional features give excellence in accuracy and real-time speed [20][21][22][23][24].…”