Mixer U-Net: An Improved Automatic Road Extraction from UAV Imagery

Abstract

Automatic road extraction from unmanned aerial vehicle (UAV) imagery has been one of the major research topics in the area of remote sensing analysis due to its importance in a wide range of applications such as urban planning, road monitoring, intelligent transportation systems, and automatic road navigation. Thanks to the recent advances in Deep Learning (DL), the tedious manual segmentation of roads can be automated. However, the majority of these models are computationally heavy and, thus, are not suitable for UAV remote-sensing tasks with limited resources. To alleviate this bottleneck, we propose two lightweight models based on depthwise separable convolutions and ConvMixer inception block. Both models take the advantage of computational efficiency of depthwise separable convolutions and multi-scale processing of inception module and combine them in an encoder–decoder architecture of U-Net. Specifically, we substitute standard convolution layers used in U-Net for ConvMixer layers. Furthermore, in order to learn images on different scales, we apply ConvMixer layer into Inception module. Finally, we incorporate pathway networks along the skip connections to minimize the semantic gap between encoder and decoder. In order to validate the performance and effectiveness of the models, we adopt Massachusetts roads dataset. One incarnation of our models is able to beat the U-Net’s performance with 10× fewer parameters, and DeepLabV3’s performance with 12× fewer parameters in terms of mean intersection over union (mIoU) metric. For further validation, we have compared our models against four baselines in total and used additional metrics such as precision (P), recall (R), and F1 score.

A Novel Levy Walk-Based Framework for Scheduling Power-Intensive Mobile Edge Computing Tasks

“…Pointwise convolutions then use 1x1 convolutions to combine the output channels from the depthwise step, effectively aggregating the information [66]. Depthwise separable convolutions significantly reduce the number of parameters and computation while maintaining model performance, making them popular in mobile and embedded applications [67].…”

Archimedes optimisation algorithm quantum dilated convolutional neural network for road extraction in remote sensing images

“…Many studies concur that the extraction of roads in aerial images becomes a difficult task because of the shadows and occlusion of trees and buildings along with the different kinds of roads in aerial images, and these conditions undoubtedly make it difficult to accurately extract roads [ 5 ]. For the extraction of roads from aerial imagery, the earlier studies learned the features and characteristics of roads and categorized them into five aspects: geometrical aspects, which include the curvature and elongation of the roads, radiometric aspects [ 6 ], i.e.…”

Road-MobileSeg: Lightweight and Accurate Road Extraction Model from Remote Sensing Images for Mobile Devices

“…The spatial attention mechanism was applied to extract road-related spatial details, whereas the channel attention mechanism was used to adjust the spectral characteristics of remote sensing images. Furthermore, Sultonov et al [ 30 ] replaced the conventional convolutional layers in U-Net with ConvMixer layers, which significantly reduced the computations of the model. In addition, Alshaikhli et al [ 31 ] combined U-Net and residual blocks and used fewer convolutional layers, achieving better prediction results than the standard U-Net model.…”