The DOI matches its Crossref record; 23 of the 27 references it cites were found in a registry or an index.
Road Extraction by Deep Residual U-Net
Open accessAbstract
Road extraction from aerial images has been a hot research topic in the field of remote sensing image analysis. In this letter, a semantic segmentation neural network, which combines the strengths of residual learning and U-Net, is proposed for road area extraction. The network is built with residual units and has similar architecture to that of U-Net. The benefits of this model are twofold: first, residual units ease training of deep networks. Second, the rich skip connections within the network could facilitate information propagation, allowing us to design networks with fewer parameters, however, better performance. We test our network on a public road data set and compare it with U-Net and other two state-of-the-art deep-learning-based road extraction methods. The proposed approach outperforms all the comparing methods, which demonstrates its superiority over recently developed state of the arts.
Overview
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“In this study, we opt for a deep residual U-Net [15], [44], which has already been successfully used for retinal OCT images.”
From Interactive Deep Learning-Based Retinal OCT Layer Segmentation Refinement by Regressing Translation Maps (2024) · cites this paper as [44]
Experimental — accuracy not yet measured
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Deep Residual Learning for Image Recognition
He, K., Zhang, X., Ren, S., & Sun, J. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.90 · Cited by 235,474 · cited 5×
“This somehow shares similar idea to that of residual neural network [21].”
BackgroundMethodResultInfluentialVery Deep Convolutional Networks for Large-Scale Image Recognition
Simonyan, K., & Zisserman, A. (2014) · arXiv · doi:10.48550/arxiv.1409.1556 · Cited by 114,148 · cited 1×
“Recently, lots of works have suggested that a deeper network would have better performance [19], [20].”
BackgroundU-Net: Convolutional Networks for Biomedical Image Segmentation
Ronneberger, O., Fischer, P., & Brox, T. (2015) · Lecture notes in computer science · doi:10.1007/978-3-319-24574-4_28 · Cited by 99,404 · cited 7×
“1) U-Net: In semantic segmentation, to get a finer result, it is very important to use low-level details, while retaining high-level semantic information [23], [24].”
BackgroundMethodInfluentialFaster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
Ren, S., He, K., Girshick, R., & Sun, J. (2017) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2016.2577031 · Cited by 73,472 · cited 1×
“of-the-art performance on a variety of computer vision tasks, such as scene recognition [13] and object detection [14].”
BackgroundFaster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
Ren, S., He, K., Girshick, R., & Sun, J. (2015) · arXiv · doi:10.48550/arxiv.1506.01497 · Cited by 70,812
Going deeper with convolutions
Szegedy, C., Liu, W., Jia, Y., Sermanet, P., Reed, S., Anguelov, D., Erhan, D., Vanhoucke, V., & Rabinovich, A. (2015) · 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2015.7298594 · Cited by 48,562 · cited 1×
“Recently, lots of works have suggested that a deeper network would have better performance [19], [20].”
BackgroundFully convolutional networks for semantic segmentation
Long, J., Shelhamer, E., & Darrell, T. (2015) · 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2015.7298965 · Cited by 38,834 · cited 3×
“One way to solve this problem is employing a pretrained network and then finetuning it on the target data set, as done in [23].”
BackgroundMethodIdentity mappings in deep residual networks
He, K., Zhang, X., Ren, S., & Sun, J. (2016) · Lecture Notes in Computer Science · doi:10.1007/978-3-319-46493-0_38 · Cited by 10,741 · cited 2×
“[22] presented a detailed discussion on impacts of different combinations and suggested a full preactivation design as shown in Fig.”
BackgroundMethodDeep Learning for Remote Sensing Data: A Technical Tutorial on the State of the Art
Zhang, L., Zhang, L., & Du, B. (2016) · IEEE Geoscience and Remote Sensing Magazine · doi:10.1109/mgrs.2016.2540798 · Cited by 2,272
Learning to Detect Roads in High-Resolution Aerial Images
Mnih, V., & Hinton, G. E. (2010) · European Conference on Computer Vision · doi:10.1007/978-3-642-15567-3_16 · Cited by 681 · cited 11×
“To demonstrate the accuracy and efficiency of the proposed deep ResUnet, we test it on Massachusetts roads data set2 and compare it with three state of the art methods, including Mnih’s [2] method, Saito’s method [5], and U-Net [24].”
BackgroundMethodResultInfluential
Research works printed but not linked: 4
- zhou, 2014, Learning deep features for scene recognition using places database, Proc NIPS, 487
- mnih, 2012, Learning to label aerial images from noisy data, Proc ICML, 567
- ehrig, 2005, Relaxed precision and recall for ontology matching, Proc Workshop Integr Ontol, 25
- chollet, 2015, Keras
Cited by
Sorted by citations
Items 201–300 of 1,001
Chen, Z., Li, D., Fan, W., Guan, H., Wang, C., & Li, J. (2021) · Remote Sensing · doi:10.3390/rs13132524 · Cited by 78
“The classical segmentation methods based on deep learning include FCN [52], PSPNet [20], U-Net [53,54] , DANet [55] and Residual U-Net [53] etc.”
MethodAdams, C., Alrashed, M., An, R., Anthony, J., Asaadi, J., Ashkenazi, A., Auger, M., Balasubramanian, S., Baller, B., Barnes, C., et al. (2018) · Physical Review D · doi:10.1103/physrevd.99.092001 · Cited by 72
Li, D., Xiao, C., Liu, Y., Chen, Z., Hassan, H., Su, L., Liu, J., Li, H., Xie, W., Zhong, W., et al. (2022) · Diagnostics · doi:10.3390/diagnostics12081788 · Cited by 94
“The intended segmentation system consists of a two-stage segmentation scheme (dependent segmentation) based on replaceable segmentation networks such as 3D U-Net [48], Res U-Net [49], SegNet [50], DeepLabV3+ [51], and UNETR [52].”
MethodNejad, J. G., Ju, M.-S., Jo, J., Oh, K., Lee, Y., Lee, S.-D., Kim, E.-J., Roh, S., & Lee, H.-G. (2024) · Animals · doi:10.3390/ani14030435 · Cited by 70
“For starters, these techniques have the potential to improve estimation accuracy and , AI approaches can improve the scalability and efficiency of estimation procedures, allowing for broader spatial and temporal scale assessments [85].”
BackgroundXue, H., Tseng, E., Knott, K. D., Kotecha, T., Brown, L., Plein, S., Fontana, M., Moon, J., & Kellman, P. (2020) · Magnetic Resonance in Medicine · doi:10.1002/mrm.28291 · Cited by 28
“ter resampling. The corresponding LV mask is given in Figure 2(a). Figure 2(b) illustrate the temporal signal with foot, peak, and valley times annotated. The U-net semantic segmentation architecture (34,35) was modified for this application of AIF LV detection. As shown in Figure 3, a ResNet module (36) was used as t…”
MethodA novel multi-attention, multi-scale 3D deep network for coronary artery segmentation
Dong, C., Xu, S.-Y., Dai, D.-W., Zhang, Y.-Z., Zhang, C., & Li, Z. (2023) · Medical Image Anal. · doi:10.1016/j.media.2023.102745 · Cited by 69
Super U-Net: a modularized generalizable architecture
Beeche, C., Singh, J., Leader, J., Gezer, S., Oruwari, A. P., Dansingani, K., Chhablani, J., & Pu, J.-T. (2022) · Pattern Recognition · doi:10.1016/j.patcog.2022.108669 · Cited by 69
Automating cell counting in fluorescent microscopy through deep learning with c-ResUnet
Morelli, R., Clissa, L., Amici, R., Cerri, M., Hitrec, T., Luppi, M., Rinaldi, L., Squarcio, F., & Zoccoli, A. (2021) · Scientific Reports · doi:10.1038/s41598-021-01929-5 · Cited by 67
“Remarkably, the careful additions with respect to the ResUnet 44 —i.e. a learned colorspace transformation and a residual block with 5 × 5 filters—enable the model to perform better than the original Unet 33 despite having seven times fewer parameters.”
MethodInfluentialZhou, X., Gong, Q., Liu, Y., & Yin, L. (2021) · Automation in Construction · doi:10.1016/j.autcon.2021.103685 · Cited by 78
Jiang, Z., Tahmasebi, P., & Mao, Z. (2021) · Advances in Water Resources · doi:10.1016/j.advwatres.2021.103878 · Cited by 69
Ghaffarian, S., Kerle, N., Pasolli, E., & Arsanjani, J. J. (2019) · Remote Sensing · doi:10.3390/rs11202427 · Cited by 69
“Furthermore, since the method The method is primarily based on the adapted deep residual U-net [54] to automatically train and detect buildings for post-disaster situations. is a pixel-level classification, it can extract the shape and size of the buildings and, thus, their changes.”
BackgroundMethodInfluentialDixit, M., Chaurasia, K., & Mishra, V. (2021) · Expert systems with applications · doi:10.1016/j.eswa.2021.115530 · Cited by 68
Ayala, C., Sesma, R., Aranda, C., & Galar, M. (2021) · Remote Sensing · doi:10.3390/rs13163135 · Cited by 64
“[43] combined the strengths of residual learning with the U-Net architecture aiming at facilitating the propagation of information while reducing the number of parameters.”
BackgroundChemGrapher: Optical Graph Recognition of Chemical Compounds by Deep Learning
Oldenhof, M., Arany, Á., Moreau, Y., & Simm, J. (2020) · Journal of Chemical Information and Modeling · doi:10.1021/acs.jcim.0c00459 · Cited by 67
Skouta, A., Elmoufidi, A., Jai-Andaloussi, S., & Ouchetto, O. (2022) · Journal of Big Data · doi:10.1186/s40537-022-00632-0 · Cited by 67
“definition techniques are used in various application areas such as: image classification [3, 11, 12], road mining [13], segmentation of brain tumors [14] and object detection [15].”
MethodCascaded Attention DenseUNet (CADUNet) for Road Extraction from Very-High-Resolution Images
Li, J., Liu, Y., Zhang, Y.-D., & Zhang, Y. (2021) · ISPRS Int. J. Geo Inf. · doi:10.3390/ijgi10050329 · Cited by 67
“[33] proposed deep residual U-Net for road extraction.”
Wang, J., Chen, S.-B., Ding, C., Tang, J., & Luo, B. (2021) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2021.3102026 · Cited by 67
“extraction [3], [4], loss estimation on pest and disease, forecast of agricultural harvest, and so on.”
BackgroundCaraNet: context axial reverse attention network for segmentation of small medical objects
Lou, A., Guan, S., Ko, H., & Loew, M. (2022) · Medical Imaging · doi:10.1117/12.2611802 · Cited by 65
“We compared CaraNet with six medical image segmentation models, including state-of-the-art models: U-Net [1], U- Net++ [2], ResUNet-mod [43], ResUNet++ [3], SFA [44], and PraNet [27].”
MethodApplication of deep neural network learning in composites design
Wang, Y., Soutis, C., Ando, D., Sutou, Y., & Narita, F. (2022) · European Journal of Materials · doi:10.1080/26889277.2022.2053302 · Cited by 69
Self-citationPham, V.-T., Tran, T.-T., Wang, P., Chen, P.-Y., & Lo, M. (2021) · Artificial Intelligence in Medicine · doi:10.1016/j.artmed.2021.102065 · Cited by 68
Real-Time Polyp Detection, Localisation and Segmentation in Colonoscopy Using Deep Learning
Jha, D., Ali, S., Johansen, H., Johansen, D., Rittscher, J., Riegler, M., & Halvorsen, P. (2020) · arXiv.org · doi:10.48550/arxiv.2011.07631 · Cited by 65
“ResUNet [56] 8,420,077 1e−4 Adam 8 Dice loss -”
MethodDeep Learning for Regular Change Detection in Ukrainian Forest Ecosystem With Sentinel-2
Isaienkov, K., Yushchuk, M., Khramtsov, V., & Seliverstov, O. (2020) · IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing · doi:10.1109/jstars.2020.3034186 · Cited by 78
“In [13], the better performance of deep learning (Sharp-Mask [47], U-Net, and ResUNet [61]) models with respect to classic machine learning (random forest [7] and multilayer perceptron [25]) algorithms to track the change detection of deforestation in the Amazon using Landsat data are presented.”
Zhang, Y., Wu, J., Liu, Y., Chen, Y.-F., Wu, E., & Tang, X.-Y. (2020) · IEEE journal of biomedical and health informatics · doi:10.1109/jbhi.2020.2996783 · Cited by 65
Satellite Imagery Multiscale Rapid Detection with Windowed Networks
Etten, A. V. (2018) · IEEE Workshop/Winter Conference on Applications of Computer Vision · doi:10.1109/wacv.2019.00083 · Cited by 65
“Efforts such as [17], [26] have shown success in extracting roads from overhead imagery via segmentation techniques.”
MethodLiu, J., Wang, X., & Tai, X. (2022) · Journal of Mathematical Imaging and Vision · doi:10.1007/s10851-022-01087-x · Cited by 53
“tly expanded in the field of medical image, such as Unet++ [25], Vnet [26], Progressive Dense V-net (PDV-Net) [27]. Based on U-Net, many networks have been developed for other segmentation tasks [28], [29]. The second architecture is to use the dilated/atrous convolution rather than the pooling layer. Dilated convoluti…”
Qi, L., Wang, M.-H., Hu, C.-M., & Holt, B. (2022) · Remote Sensing of Environment · doi:10.1016/j.rse.2022.113188 · Cited by 63
Gaggion, N., Ariel, F., Daric, V., Lambert, É., Legendre, S., Roulé, T., Camoirano, A., Milone, D. H., Crespi, M. D., Blein, T., et al. (2021) · GigaScience · doi:10.1093/gigascience/giab052 · Cited by 92
“In the first model, named DSResUNet, we aimed at improving the segmentation accuracy while keeping at the same time a fast lightweight model.”
InfluentialTopology-Enhanced Urban Road Extraction via a Geographic Feature-Enhanced Network
Li, X.-G., Wang, Y.-B., Zhang, L., Liu, S.-H., Mei, J., & Li, Y. (2020) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2020.2991006 · Cited by 63
“The method with pixel-to-pixel work-flow includes the fully convolutional networks (FCNs) [25], Unet [26], Unet ++ [27], D-Linknet [5], SegNet [2], and ResUnet [4].”
BackgroundMethodInfluentialSelf-citationEvolutionary Compression of Deep Neural Networks for Biomedical Image Segmentation
Zhou, Y., Yen, G. G., & Yi, Z. (2020) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2019.2933879 · Cited by 62
Deep Semantic Segmentation of Center Pivot Irrigation Systems from Remotely Sensed Data
de Albuquerque, A. O., de Carvalho, O. D., de Carvalho, O. L. F., de Bem, P. P., Ferreira, P. H. G., dos Santos, D. M. R., Silva, C. R. E., Gomes, R., & Guimarães, R. (2020) · Remote Sensing · doi:10.3390/rs12132159 · Cited by 61
“In this present research, we used three deep learning architectures: U-net [76], Deep ResUnet [77], and SharpMask [78].”
BackgroundMethodResultInfluentialDeep learning for dune pattern mapping with the AW3D30 global surface model
Shumack, S., Hesse, P., & Farebrother, W. (2020) · Earth Surface Processes and Landforms · doi:10.1002/esp.4888 · Cited by 61
“U-Net, discussed in detail in the Materials and Methods section, was originally developed for medical image segmentation (Ronneberger et al ., 2015), but has since been adapted to remote sensing applications (e.g. Zhang et al ., 2018) and landform classification (e.g. Li et al ., 2020).”
MethodOcclusion-Aware Road Extraction Network for High-Resolution Remote Sensing Imagery
Yang, R., Zhong, Y., Liu, Y., Lu, X., & Zhang, L. (2024) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2024.3387945 · Cited by 52
“…algorithms that focus on hand-crafted features to extract roads [14], [15], [16], deep neural network models [17], [18], [19] have achieved a satisfactory accuracy in the automatic extraction of roads from remote sensing imagery [20], [21], especially with the open road dataset [7], [22], [23].”
BackgroundMethodProgressive Multiscale Consistent Network for Multiclass Fundus Lesion Segmentation
He, A., Wang, K., Li, T., Wang, B., Kang, H., & Fu, H. (2022) · IEEE Transactions on Medical Imaging · doi:10.1109/tmi.2022.3177803 · Cited by 77
“UNet-based networks include UNet [33], UNet++ [48], ResUNet [46], Att-UNet [29] and DenseUNet [25].”
InfluentialSu, Z., Wang, Y., Xu, Q., Gao, R., & Kong, Q.-M. (2022) · Computers and Electronics in Agriculture · doi:10.1016/j.compag.2022.106873 · Cited by 60
Self-citationMelanoma segmentation using deep learning with test-time augmentations and conditional random fields
Ashraf, H., Waris, A., Ghafoor, M. F., Gilani, S., & Niazi, I. (2022) · Scientific Reports · doi:10.1038/s41598-022-07885-y · Cited by 60
“For the first time, we investigate how to utilize the discriminative capacity of UNet-based three separate semantic segmentation deep neural network architectures: UNet 15 , deep residual U-Net (ResUNet) 16 , and improved ResUNet (ResUNet++) 17 at the model training level.”
MethodInfluentialAutomatic Coal and Gangue Segmentation Using U-Net Based Fully Convolutional Networks
Gao, R., Sun, Z., Li, W., Pei, L., Hu, Y., & Xiao, L. (2020) · Energies · doi:10.3390/en13040829 · Cited by 64
Road Extraction from Very High Resolution Images Using Weakly labeled OpenStreetMap Centerline
Wu, S., Du, C., Chen, H., Xu, Y., Guo, N., & Jing, N. (2019) · ISPRS International Journal of Geo-Information · doi:10.3390/ijgi8110478 · Cited by 64
“The ResUNet was first proposed in the [4], and D-linknet was the winner of the CVPR2018 digital challenge of road extraction.”
BackgroundMethodInfluentialMK-UNet: Multi-Kernel Lightweight CNN for Medical Image Segmentation
Rahman, M. M., & Mărculescu, R. (2025) · IEEE International Conference on Computer Vision Workshops · doi:10.1109/iccvw69036.2025.00114 · Cited by 52
“Inspired by residual learning in ResNet [13], ResUNet [43] uses residual blocks to facilitate gradient flow and improve convergence, addressing the vanishing gradient problem in deep networks.”
Single-cell spatial transcriptomic atlas of the whole mouse brain
Han, L., Liu, Z., Jing, Z., Liu, Y., Chang, H., Lei, J., Peng, Y., Wang, K., Xu, Y., Liu, W., et al. (2025) · Neuron · doi:10.1016/j.neuron.2025.02.015 · Cited by 77
Chowdary, J., Yogarajah, P., Chaurasia, P., & Guruviah, V. (2022) · Ultrasonic imaging (Print) · doi:10.1177/01617346221075769 · Cited by 59
“42 The architecture of the proposed residual module is different from Diakogiannis et al. 43 , Zhang et al. 44 in terms of number of layers. this proposed work used three convolutions, three batch-normalization, three ReLU, and one concatenation layer.”
MethodLiver segmentation from computed tomography images using cascade deep learning
Araújo, J. D. L., da Cruz, L. B., Diniz, J. O. B., Ferreira, J. L., Silva, A. C., de Paiva, A. C., & Gattass, M. (2021) · Computers in Biology and Medicine · doi:10.1016/j.compbiomed.2021.105095 · Cited by 77
Wu, H., Chen, G., Wen, Z.-K., & Qin, J. (2021) · IEEE International Conference on Computer Vision · doi:10.1109/iccv48922.2021.00347 · Cited by 59
“…superiority of our proposed method, we compared it with seven state-of-the-arts meth-ods on Kvasir-SEG and CVC-Clinic DB, including six fully supervised networks (ResU-Net[37], U-Net++[40], CE-Net[10], CPF-Net[8], PraNet[7] and HarDNet-MSEG[12]) and one semi-supervised network (Hung’s method [13]).”
MethodLandmark Detection in Cardiac MRI by Using a Convolutional Neural Network
Xue, H., Artico, J., Fontana, M., Moon, J., Davies, R. H., & Kellman, P. (2021) · Radiology Artificial Intelligence · doi:10.1148/ryai.2021200197 · Cited by 65
“A variation of U-Net architecture was implemented (25,26) for heatmap detection.”
MethodAdvanced Fully Convolutional Networks for Agricultural Field Boundary Detection
Taravat, A., Wagner, M. P., Bonifaçio, R., & Petit, D. (2021) · Remote Sensing · doi:10.3390/rs13040722 · Cited by 59
“The main advantage of using the U-Net architecture is that it works with very few training images and yields more precise segmentation [43].”
Zhang, X., Li, J., & Hua, Z. (2022) · IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing · doi:10.1109/jstars.2022.3185245 · Cited by 66
“In addition, these encoder–decoder-based network structures have been widely used in other directions in remote sensings, such as road segmentation [57], [58], classification [59], [60], and building detection [61], [62].”
MethodReal-Time Ultrasound Image Despeckling Using Mixed-Attention Mechanism Based Residual UNet
Lan, Y., & Zhang, X.-M. (2020) · IEEE Access · doi:10.1109/access.2020.3034230 · Cited by 59
“1(b), which is a deep residual UNet [33], [34] with mixed-attention mechanism.”
BackgroundKim, D.-K., Kim, Y., Kim, H., & Kim, D. (2022) · IEEE Journal of Biomedical and Health Informatics · doi:10.1109/jbhi.2022.3172514 · Cited by 69
“ResUNet has the advantages of facilitating network training and accelerating network convergence by avoiding the vanishing gradient problem [43].”
BackgroundMethodAutomatic Target Detection from Satellite Imagery Using Machine Learning
Tahir, A., Munawar, H. S., Akram, J., Adil, M., Ali, S., Kouzani, A., & Mahmud, M. A. P. (2022) · Italian National Conference on Sensors · doi:10.3390/s22031147 · Cited by 58
Li, Z., Xin, Q., Sun, Y., & Cao, M.-Y. (2021) · Remote Sensing · doi:10.3390/rs13183630 · Cited by 58
“U-Net, originally proposed for biomedical image processing and implemented in Caffe [40] was found to be effective in the semantic segmentation of remote sensing images [41,42].”
BackgroundMethodGuo, Y., Liao, J.-J., & Shen, G. (2021) · Remote Sensing · doi:10.3390/rs13020245 · Cited by 57
“U-net has an encoder–decoder structure that is widely used in remote sensing image classification due to its high flexibility and superiority [67].”
MethodAn Effective Lunar Crater Recognition Algorithm Based on Convolutional Neural Network
Wang, S., Fan, Z., Li, Z.-M., Zhang, H., & Wei, C. (2020) · Remote Sensing · doi:10.3390/rs12172694 · Cited by 57
“Then, we use Cartopy [24], which is a Python package to transform the image into an orthographic projection, in order to minimize image distortion.”
MethodAutomated Inline Analysis of Myocardial Perfusion MRI with Deep Learning
Xue, H., Davies, R. H., Brown, L., Knott, K. D., Kotecha, T., Fontana, M., Plein, S., Moon, J., & Kellman, P. (2020) · Radiology Artificial Intelligence · doi:10.1148/ryai.2020200009 · Cited by 62
“data. Neural Net model The first 48 images were empirically selected, resulting in an image array of 176×176×48, which was used for training and testing. The U-net semantic segmentation architecture [23,24] was adopted for the perfusion segmentation. As shown in Figure 2, the neural net consists of downsample and upsa…”
MethodWu, Q., Luo, F., Wu, P., Wang, B., Yang, H., & Wu, Y. (2020) · IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing · doi:10.1109/jstars.2020.3042816 · Cited by 71
“[37] combined ResNet and U-net [38] to extract road networks, which are designed with few parameters but achieve good performance.”
BackgroundMulti-Object Segmentation in Complex Urban Scenes from High-Resolution Remote Sensing Data
Abdollahi, A., Pradhan, B., Shukla, N., Chakraborty, S., & Alamri, A. (2021) · Remote Sensing · doi:10.3390/rs13183710 · Cited by 57
“We compared the results of our methods for both roads and buildings with other comparative methods, such as Res-U-Net [64], JointNet [65], DeeplabV3 [58], and LinkNet [68].”
MethodResultUNet-Like Remote Sensing Change Detection: A review of current models and research directions
Wu, C., Zhang, L.-P., Du, B., Chen, H., Wang, J., & Zhong, H. (2024) · IEEE Geoscience and Remote Sensing Magazine · doi:10.1109/mgrs.2024.3412770 · Cited by 56
“As a result, UNet has become the most widely used change detection model in current research [60], [61].”
Towards Vision-Language Geo-Foundation Model: A Survey
Zhou, Y., Feng, L., Ke, Y.-P., Jiang, X., Yan, J.-C., Yang, X., & Zhang, W. (2024) · Photogrammetric Record · doi:10.48550/arxiv.2406.09385 · Cited by 56
“…advancements in nearly all geospatial tasks, such as scene classification [1], object detection [2], [3], change detection [4], denoising [5], land use segmentation [6], disaster management [7], and geospatial localization [8], driven by deep learning and other artificial intelligence technologies.”
Deep learning downscaled high-resolution daily near surface meteorological datasets over East Asia
Lin, H., Tang, J., Wang, S., Wang, S., & Dong, G. (2023) · Scientific Data · doi:10.1038/s41597-023-02805-9 · Cited by 62
The Sound Demixing Challenge 2023 - Music Demixing Track
Fabbro, G., Uhlich, S., Lai, C.-H., Choi, W., Martínez-Ramírez, M. A., Liao, W.-H., Gadelha, I., Ramos, G. F., Hsu, E., Rodrigues, H., et al. (2023) · Transactions of the International Society for Music Information Retrieval · doi:10.5334/tismir.171 · Cited by 56
Khankeshizadeh, E., Mohammadzadeh, A., Moghimi, A., & Mohsenifar, A. (2022) · Earth Science Informatics · doi:10.1007/s12145-022-00885-6 · Cited by 56
“For instance, in (Drusch et al. 2012), deep learning models (e.g., SharpMask (Pinheiro et al. 2016), U-Net (Ron-neberger et al. 2015), and ResU-Net (Zhang et al. 2018)) revealed better performance than the classic machine learning models (e.g., random forest (Breiman 2001) and multi-layer…”
BackgroundResultInfluentialGraph Convolution Based Cross-Network Multiscale Feature Fusion for Deep Vessel Segmentation
Zhao, G., Liang, K., Pan, C., Zhang, F., Wu, X., Hu, X., & Yu, Y. (2022) · IEEE Transactions on Medical Imaging · doi:10.1109/tmi.2022.3207093 · Cited by 63
“The methods in these comparisons include DDT [10], DVS [8], UNET3d [20], nnUNET [27], ResUNET [40], DenseUNET [21], PSP-Net [17] and HMSA [18].”
MethodTransResU-Net: Transformer based ResU-Net for Real-Time Colonoscopy Polyp Segmentation
Tomar, N. K., Shergill, A., Rieders, B., Bagci, U., & Jha, D. (2022) · arXiv.org · doi:10.48550/arxiv.2206.08985 · Cited by 56
“One of the promising extensions of UNet is Re-sUNet [20].”
MethodBongratz, F., Rickmann, A.-M., Pölsterl, S., & Wachinger, C. (2022) · Computer Vision and Pattern Recognition · doi:10.1109/cvpr52688.2022.02011 · Cited by 56
“Image encoding and decoding (CNN) We use a residual UNet architecture [3, 22, 42, 56] for image-feature extraction.”
MethodExtreme Rotation Estimation using Dense Correlation Volumes
Cai, R., Hariharan, B., Snavely, N., & Averbuch‐Elor, H. (2021) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr46437.2021.01433 · Cited by 33
“We adopt a DeepResidual-Unet(ResUNet) [11] architecture as the sharedweight Siamese encoder, which downsamples and upsamples the input image and outputs 32×H/4×W/4 embedded feature maps.”
MethodShip Detection Using a Fully Convolutional Network with Compact Polarimetric SAR Images
Fan, Q., Chen, F., Cheng, M., Lou, S., Xiao, R., Zhang, B., Wang, C., & Li, J. (2019) · Remote Sensing · doi:10.3390/rs11182171 · Cited by 56
“Currently, U-Net is used in crop mapping with SAR data [37] and road extraction with optical images [38].”
MethodA Novel Multi-Branch Channel Expansion Network for Garbage Image Classification
Shi, C.-P., Xia, R.-Y., & Wang, L.-G. (2020) · IEEE Access · doi:10.1109/access.2020.3016116 · Cited by 56
“However, when the network is deep enough, further deepening the network will not improve the network performance but make the network more difficult to train and even make it performance decline [45].”
BackgroundA flexible deep learning crater detection scheme using Segment Anything Model (SAM)
Giannakis, I., Bhardwaj, A., Sam, L., & Leontidis, G. (2023) · Icarus · doi:10.1016/j.icarus.2023.115797 · Cited by 75
“…standard in remote sensing segmentation in geosciences (Buscombe and Goldstein, 2022; Chen et al., 2020; Collins et al., 2020; Gupta et al., 2021; Zhang et al., 2018), and for real time identification of objects in Martian terrain for safe rover navigation (Liu et al., 2023a; Goh et al., 2022;…”
BackgroundCombining ICESat-2 photons and Google Earth Satellite images for building height extraction
Zhao, Y., Wu, B., Li, Q.-X., Yang, L., Fan, H.-C., Wu, J.-P., & Yu, B.-L. (2023) · International Journal of Applied Earth Observation and Geoinformation · doi:10.1016/j.jag.2023.103213 · Cited by 55
Ali, S., Espinel, Y., Jin, Y., Liu, P., Güttner, B., Zhang, X.-K., Zhang, L., Dowrick, T., Clarkson, M. J., Xiao, S., et al. (2024) · Medical Image Anal. · doi:10.48550/arxiv.2401.15753 · Cited by 54
“One ResUnet segmented the ridge landmark from the original image, and the other one segmented the silhouette and the ligament landmarks from the contour-enhanced image.”
MethodLearning-based Localizability Estimation for Robust LiDAR Localization
Nubert, J., Walther, E., Khattak, S., & Hutter, M. (2022) · IEEE/RJS International Conference on Intelligent RObots and Systems · doi:10.1109/iros47612.2022.9982257 · Cited by 54
“Finally, the performance of the proposed architecture is superior on the test set; when considering F1 score, precision and recall, the practical advantage of ResUNet over PointNet is underlined.”
InfluentialJeon, E.-I., Kim, S., Park, S., Kwak, J., & Choi, I. (2021) · Ecological Informatics · doi:10.1016/j.ecoinf.2021.101430 · Cited by 62
Chen, Z., Wang, C., Li, J., Fan, W., Du, J., & Zhong, B. (2021) · International Journal of Applied Earth Observation and Geoinformation · doi:10.1016/j.jag.2021.102341 · Cited by 58
“Area extraction and centerline extraction are the two major aspects of works for road extraction from remote sensing images (Cheng et al., 2017a, 2017b, Zhang et al. 2017).”
BackgroundMethodInfluentialDeep learning assisted Shack-Hartmann wavefront sensor for direct wavefront detection.
Hu, L.-J., Hu, S., Gong, W., & Si, K. (2020) · Optics Letters · doi:10.1364/ol.395579 · Cited by 54
Thermal Infrared Pedestrian Segmentation Based on Conditional GAN
Wang, P., & Bai, X.-Z. (2019) · IEEE Transactions on Image Processing · doi:10.1109/tip.2019.2924171 · Cited by 54
Zhu, Z., Wang, Z., Qi, G., Zhao, Y., & Liu, Y. (2025) · IEEE Transactions on Instrumentation and Measurement · doi:10.1109/tim.2025.3551581 · Cited by 25
“Many variants of U-Net, such as UNet ++ [17], Res-Unet [18], and X-net [19], have further extended the research on U-Net.”
DeepWindow: Sliding Window Based on Deep Learning for Road Extraction From Remote Sensing Images
Lian, R., & Huang, L. (2020) · IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing · doi:10.1109/jstars.2020.2983788 · Cited by 65
“[17] and Alexander et al .”
Detection of Pavement Cracks by Deep Learning Models of Transformer and UNet
Zhang, Y., & Zhang, L. (2024) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2024.3420763 · Cited by 54
“…modifications in the architecture of UNet itself, theoretically, various end-to-end networks for image segmentation tasks, such as UNet ++ [8], [43], Deep Residual UNet (ResUNet) [7], ResUNet ++ [10], Attention UNet (AttnUNet) [9], etc., can be used for crack detection through transfer learning.”
MethodVision based Crop Row Navigation under Varying Field Conditions in Arable Fields
de Silva, R., Cielniak, G., & Gao, J.-F. (2022) · Computers and Electronics in Agriculture · doi:10.48550/arxiv.2209.14003 · Cited by 53
Lin, D., Li, Y.-Q., Nwe, T., Dong, S., & Oo, Z. (2020) · Pattern Recognition Letters · doi:10.1016/j.patrec.2020.07.013 · Cited by 53
SiameseGAN: A Generative Model for Denoising of Spectral Domain Optical Coherence Tomography Images
Kande, N. A., Dakhane, R., Dukkipati, A., & Yalavarthy, P. K. (2020) · IEEE Transactions on Medical Imaging · doi:10.1109/tmi.2020.3024097 · Cited by 76
“2) Deep Residual U-Net Architecture: The U-Net architecture [25] also uses skip connection to facilitate the training of the network.”
BackgroundMethodDSU-Net: Dual-Stage U-Net based on CNN and Transformer for skin lesion segmentation
Zhong, L., Li, T.-S., Cui, M., Cui, S.-G., Wang, H.-K., & Yu, L. (2025) · Biomedical Signal Processing and Control · doi:10.1016/j.bspc.2024.107090 · Cited by 52
Pan, D., Shen, J., Al-Huda, Z., & Al-qaness, M. (2025) · Comput. Biol. Medicine · doi:10.1016/j.compbiomed.2025.109662 · Cited by 52
Generating synthetic computed tomography for radiotherapy: SynthRAD2023 challenge report
Huijben, E. M. C., Terpstra, M. L., Galapon, A., Pai, S., Thummerer, A., Koopmans, P. J., Afonso, M. V., van Eijnatten, M. A. J. M., Gurney‐Champion, O. J., Chen, Z., et al. (2024) · Medical Image Analysis · doi:10.1016/j.media.2024.103276 · Cited by 93
Self-citationNext-Gen Medical Imaging: U-Net Evolution and the Rise of Transformers
Zhang, C. H., Deng, X., & Ling, S. H. (2024) · Sensors · doi:10.3390/s24144668 · Cited by 64
“ResUNet (Residual and U-Net), introduced by [27], is a typical example of combining the residual network and U-Net.”
BackgroundAutomatic Windthrow Detection Using Very-High-Resolution Satellite Imagery and Deep Learning
Kislov, D. E., & Korznikov, K. (2020) · Remote Sensing · doi:10.3390/rs12071145 · Cited by 67
“[43] proposed a road segmentation algorithm based on U-Net.”
MethodLi, Y., Xu, L., Rao, J., Guo, L., Yan, Z., & Jin, S. (2019) · Remote Sensing Letters · doi:10.1080/2150704x.2018.1557791 · Cited by 52
“Shelhamer et al. and Zhang et al. proposed fully convolutional network (FCN, Shelhamer, Long, and Darrell 2017) and U-Net (Ronneberger, Fischer, and Brox 2015; Zhang, Liu, and Wang 2018) for object segmentation.”
BackgroundPotential of SDGSAT-1 nighttime light data in extracting urban main roads
Wu, B., Wang, Y., Huang, H.-L., Liu, S., & Yu, B.-L. (2024) · Remote Sensing of Environment · doi:10.1016/j.rse.2024.114448 · Cited by 51
Self-citationFan, X., Zhou, J., Jiang, X., Xin, M., & Hou, L. (2024) · Comput. Biol. Medicine · doi:10.1016/j.compbiomed.2024.108265 · Cited by 51
Zeng, Y.-F., Guo, Y., & Li, J. (2021) · Neural computing & applications (Print) · doi:10.1007/s00521-021-06027-1 · Cited by 51
Photoacoustic microscopy with sparse data by convolutional neural networks
Zhou, J., He, D., Shang, X., Guo, Z., Chen, S., & Luo, J. (2021) · Photoacoustics · doi:10.1016/j.pacs.2021.100242 · Cited by 60
“4, besides our restoration method, three other representative methods, the conventional bicubic interpolation, a re-trained EDSR model [39], and a re-trained Residual U-Net model [40], were applied for comparison.”
MethodComparing U-Net Based Models for Denoising Color Images
Komatsu, R., & Gonsalves, T. (2020) · Applied Informatics · doi:10.3390/ai1040029 · Cited by 51
“[57] constructed U-Net with residual block as Residual U-Net and extracted the roads from aerial maps.”
MethodHsgNet: A Road Extraction Network Based on Global Perception of High-Order Spatial Information
Xie, Y., Miao, F., Zhou, K., & Peng, J. (2019) · ISPRS International Journal of Geo-Information · doi:10.3390/ijgi8120571 · Cited by 58
“This paper addresses the problem of extracting road regions [1,2] from remote sensing images.”
BackgroundArtificial Intelligence for Automatic Measurement of Sagittal Vertical Axis Using ResUNet Framework
Weng, C.-H., Wang, C.-L., Huang, Y.-J., Yeh, Y.-C., Fu, C.-J., Yeh, C.-Y., & Tsai, T.-T. (2019) · Journal of Clinical Medicine · doi:10.3390/jcm8111826 · Cited by 51
“ResUNet has been used for cell segmentation from biomedical images [12] and road extraction from satellite images [15].”
MethodLiu, G., Chen, Z., Liu, D., Chang, B., & Dou, Z. (2023) · IEEE Transactions on Instrumentation and Measurement · doi:10.1109/tim.2023.3293880 · Cited by 40
“Influenced by U-Net, many variant segmentation models have been proposed, such as UNet ++ [26], UNet3 + [27], residual U-Net (ResUNet) [28], ResUNet ++ [29], adaptive context selection network (ACSNet) [30], parallel reverse attention network (PraNet) [31], and context axial reverse attention…”
MethodChen, Z.-X., Luo, Y., Wang, J., Li, J., Wang, C., & Li, D.-L. (2023) · International Journal of Applied Earth Observation and Geoinformation · doi:10.1016/j.jag.2023.103510 · Cited by 49
Automatic segmentation of large-scale CT image datasets for detailed body composition analysis
Ahmad, N., Strand, R., Sparresäter, B., Tarai, S., Lundström, E., Bergström, G., Ahlström, H., & Kullberg, J. (2023) · BMC Bioinformatics · doi:10.1186/s12859-023-05462-2 · Cited by 49
“To enhance the performance of the UNET architecture, a ResUNET model [22] was proposed, in which the traditional convolutional blocks are substituted with residual blocks.”
MethodB-FGC-Net: A Building Extraction Network from High Resolution Remote Sensing Imagery
Wang, Y., Zeng, X., Liao, X., & Zhuang, D. (2022) · Remote Sensing · doi:10.3390/rs14020269 · Cited by 56
“This scheme not only speeds up the training of the model but also effectively facilitates the utilization of low level features [39].”
BackgroundSelf-citationLu, X., Zhong, Y., Zheng, Z., & Wang, J. (2021) · ISPRS Journal of Photogrammetry and Remote Sensing · doi:10.1016/j.isprsjprs.2021.08.018 · Cited by 46
Huerta-García, R. E., Yépez-Rincón, F. D., Lozano-García, D. F., Guerra-Cobián, V. H., Fierro, A. L. F., Gómez, H. D. L., González, R. A. C., & Vargas‐Martínez, A. (2021) · Remote Sensing · doi:10.3390/rs13112031 · Cited by 59
“These types of networks have achieved wide success with state-of-the-art results for a wide variety of problems from medical applications [45,46] to their employment in remote sensing for road [47] and building extractions [48], as well as land cover classification [49], but they have not been used to make many advanc…”
BackgroundA Self-Supervised Approach to Pixel-Level Change Detection in Bi-Temporal RS Images
Chen, Y.-X., & Bruzzone, L. (2022) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2022.3203897 · Cited by 49
“We adopt a ResUnet architecture similar to the one presented in the [34] but replace all padding types with the same padding.”
BackgroundMethodSegmenting the Inferior Alveolar Canal in CBCTs Volumes: The ToothFairy Challenge
Bolelli, F., Lumetti, L., Vinayahalingam, S., Bartolomeo, M. D., Pellacani, A., Marchesini, K., van Nistelrooij, N., van Lierop, P., Xi, T., Liu, Y.-S., et al. (2024) · IEEE Transactions on Medical Imaging · doi:10.1109/tmi.2024.3523096 · Cited by 48
“A 3D ResUNet-based [38] neural network is employed, taking the full re-sampled volumes as input and producing segmentations with the same shape.”
Metadata
- Authors
- Zhang, Z., Liu, Q., & Wang, Y.
- Affiliations
- State Key Laboratory of Virtual Reality Technology and Systems (CN)
- Venue
- IEEE Geoscience and Remote Sensing Letters
- Published
- 2018-05
- Volume
- 15
- Issue
- 5
- Pages
- 749-753
- ISSN
- 1545-598X (ISSN-L)
- 1558-0571
- OpenAlex
- W2774320778
- Cited by
- 3,179
- References
- 27
- Citations by year
- 2026: 331 · 2025: 526 · 2024: 616 · 2023: 510 · 2022: 441 · 2021: 342 · 2020: 241 · 2019: 147 · 2018: 25
- FWCI
- 151.19
- Type
- journal article
- Language
- en
- Open access
- green
- arxiv.org
- Locations
- IEEE Geoscience and Remote Sensing Letters (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University) (cc-by)
- Funders
- Natural Science Foundation of China (61601011)
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Automated Road and Building Extraction
- Physical Sciences › Engineering › Civil and Structural Engineering › Infrastructure Maintenance and Monitoring
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
- Keywords
- road extraction, aerial images, semantic segmentation, residual learning, U-Net, skip connections
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
