Aerial Image Semantic Segmentation Using Neural Search Network Architecture

Bui, Duc-Thinh; Tran, Trong-Dat; Nguyen, Thi-Thuy; Tran, Quoc-Long; Nguyen, Do Van

doi:10.1007/978-3-030-03014-8_10

Cited by 7 publications

(1 citation statement)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This method can simplify the process of designing new network architectures. In remote sensing field, it also has a wide utilization (Chen et al, 2019, Bui et al, 2018. In this paper, enlightened by this automated manner, we rethink the effect of the popular "add module" pattern and focus on an automated and lightweight network design procedure in remote sensing image scene classification task.…”

Section: Introductionmentioning

confidence: 99%

Automated and Lightweight Network Design via Random Search for Remote Sensing Image Scene Classification

Diao

Sun

et al. 2020

Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci.

View full text Add to dashboard Cite

Abstract. With the development of deep learning, remote sensing image scene classification technology has been greatly improved. However, current deep networks used for scene classification usually introduce ingenious extra modules to fit the characteristics of remote sensing images. It causes a high labor cost and brings more parameters, which makes the network more complicated and poses new intractable problems. In this paper, we rethink this popular “add module” pattern and propose a more lightweight model, called ProbDenseNet (PDN). PDN is obtained via a random search strategy in Neural Architecture Search (NAS) which is an automated network design manner. In our method, all topological connections are assigned importance degrees which subject to a uniform distribution. And we set a regulator to adjust the sparsity of the network. By this way, the design procedure is more automated and the network structure becomes more lightweight. Experimental results on AID benchmark demonstrate that the proposed PDN model can achieve competitive performance even with much fewer parameters. And we also find that excessive connections do not always improve the network’s performance while they can drag down the network’s behavior as well. Furthermore, we conduct experiments on Vaihingen dataset with classical Fully Convolutional Network (FCN) framework. Quantitative and qualitative results both indicate that the features learned by PDN can also transfer in semantic segmentation task.

show abstract

Section: Introductionmentioning

confidence: 99%

Automated and Lightweight Network Design via Random Search for Remote Sensing Image Scene Classification

Diao

Sun

et al. 2020

Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci.

View full text Add to dashboard Cite

show abstract

Evolutionary NAS for aerial image segmentation with gene expression programming of cellular encoding

Broni-Bediako

Murata

Mormille

et al. 2021

Neural Comput & Applic

View full text Add to dashboard Cite

A Comprehensive Analysis of Deep Learning-Based Approaches for Prediction and Prognosis of Infectious Diseases

Thakur

Kaur

Kumar

2023

Arch Computat Methods Eng

View full text Add to dashboard Cite

Artificial intelligence is the most powerful and promising tool for the present analytic technologies. It can provide real-time insights into disease spread and predict new pandemic epicenters by processing massive amount of data. The main aim of the paper is to detect and classify multiple infectious diseases using deep learning models. The work is conducted by using 29,252 images of COVID-19, Middle East Respiratory Syndrome Coronavirus, Pneumonia, normal, Severe Acute Respiratory Syndrome, tuberculosis, viral pneumonia, and lung opacity which has been collected from various disease datasets. These datasets are used to train the deep learning models such as EfficientNetB0, EfficientNetB1, EfficientNetB2, EfficientNetB3, NASNetLarge, DenseNet169, ResNet152V2, and InceptionResNetV2. The images have been initially graphically represented using exploratory data analysis to study the pixel intensity and find anomalies by extracting the color channels in an RGB histogram. Later, the dataset has been pre-processed to remove noisy signals using image augmentation and contrast enhancement techniques. Further, feature extraction techniques such as morphological values of contour features and Otsu thresholding have been applied to extract the feature. The models have been evaluated on the basis of various parameters, and it has been discovered that during the testing phase, the InceptionResNetV2 model generated the highest accuracy of 88%, best loss value of 0.399, and root mean square error of 0.63.

show abstract

Aerial Image Semantic Segmentation Using Neural Search Network Architecture

Cited by 7 publications

References 22 publications

Automated and Lightweight Network Design via Random Search for Remote Sensing Image Scene Classification

Automated and Lightweight Network Design via Random Search for Remote Sensing Image Scene Classification

Evolutionary NAS for aerial image segmentation with gene expression programming of cellular encoding

A Comprehensive Analysis of Deep Learning-Based Approaches for Prediction and Prognosis of Infectious Diseases

Contact Info

Product

Resources

About