Investment of Classic Deep CNNs and SVM for Classifying Remote Sensing Images

Abstract

Feature extraction is an important process in image classification for achieving an efficient accuracy for the classification learning models. One of these methods is using the convolution neural networks. The use of the trained classic deep convolution neural networks as features extraction gives a considerable results in the remote sensing images classification models. So, this paper proposes three classification approaches using the support vector machine where based on the use of the ImageNet pre-trained weights classic deep convolution neural networks as features extraction from the remote sensing images. There are three convolution models that used in this paper; the Densenet 169, the VGG 16, and the ResNet 50 models. A comparative study is done by extract features using the outputs of the mentioned ImageNet pre-trained weights convolution models after transfer learning, and then use these extracted features as input features for the support vector machine classifier. The used datasets in this paper are the UC Merced land use dataset and the SIRI-WHU dataset. The comparison is based on calculating the overall accuracy to assess the classification model performance.

Efficient Classification of Remote Sensing Images Using Two Convolution Channels and SVM

“…This section illustrates the proposed methods' experiments setup and results and compares these proposed methods with previous methods' results in [20] and [21]. This comparison is established according to the OA measurements for each model.…”

“…Fig. 11 offers the comparison based on the OA calculations for the proposed model that utilizes the SVM classifier and other models mentioned in [21] with both datasets. The DenseNet 169…”

“…Transfer learning was done for these networks, and then fully connected (FC) layers were created and retrained. AlAfandy et al [21] proposed three remote sensing images' classifiers based on classic networks and SVM. They utilized three classic models as features extraction; the VGG 16, the DenseNet 169, and the ResNet 50 models.…”

“…Several researchers have suggested solutions to this issue, but they rely only on machine learning or deep learning. Not many have suggested crossbred approaches from a mixture of deep learning and machine learning, such as [9,12,21].…”

“…The SIRI-WHU dataset and the UC Merced land use dataset are utilized for training the proposed models in this paper. There are two steps to assess the efficiency of these proposals; the first step is comparing the experimental results of these proposed methods with the antecedent results in [20,21]. This comparative is done by measuring the overall accuracy (OA) for each classifier.…”