U-Net Convolutional Networks for Mining Land Cover Classification Based on High-Resolution UAV Imagery

Giang, Tuan Linh; Dang, Kinh Bac; Le, Quang Toan; Vu, Nguyen; Tong, Si Son; Pham, Van-Manh

doi:10.1109/access.2020.3030112

Cited by 60 publications

(38 citation statements)

References 59 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The semantic segmentation detection architecture used in the experiment is U‐Net. Inspired by Asma, Abdelhamid, and Youyou (2020), Giang et al (2020), and Upadhyay, Agrawal, and Vashist (2021), which evidenced the robustness and efficiency of U‐Net architecture in a small amount of data. Typically, this semantic segmentation task predicts the target objects by highlighting them in contour maps and produces corresponding pixel‐wise accuracy scores.…”

Section: Proposed Frameworkmentioning

confidence: 93%

Find the centroid: A vision‐based approach for optimal object grasping

Liong

Tan

et al. 2021

J Food Process Engineering

View full text Add to dashboard Cite

Recently, the booming development of deep learning techniques facilitates remarkable progress in a wide range of applications, has provided insightful cues in today's modern industry and research societies. For instance, the manufacturing and agricultural industries have witnessed the unparallel dominance in AI vision to integrate the production management. Therefore, owing to its impressive learning capability, this article aims to develop an automated system that can integrate data processing and packaging operations. Notably, the inspection system can capture the geometric properties of an ellipsoidal ham and thus can be grasped by the robotic arm. Prior to the centroid detection, a series of instant segmentation procedures is introduced to determine the position of the object, which incorporates both the object localization and semantic segmentation processes. As a result, the proposed model yields a promising segmentation accuracy when experimented on the ham dataset that consists of 293 images. The best Sørensen–Dice coefficient achieved is 91%, which indicates the compelling performance. To further verify the effectiveness of the proposed framework, quantitative and qualitative visualization are presented to demonstrate the quality of the image segmentation task. The proposed algorithm and the raw data are publicly available at https://github.com/jason2303779/Ham-Centroid. Practical Applications Recently, the booming development of deep learning techniques facilitates remarkable progress in a wide range of applications has provided insightful cues in today's modern industry and research societies. For instance, the manufacturing and agricultural industries have witnessed the unparallel dominance in AI vision to integrate production management. Therefore, this paper aims to develop an automated system that can integrate data processing and packaging operations due to its impressive learning capability. In brief, this paper introduces the first plausible mechanism to precisely identify the geometric center of an irregular‐shaped object, viz, ellipsoidal ham. A novel pipeline that composes both the object detection and semantic segmentation architectures is proposed to facilitate the vision understanding process. Succinctly, the inspection system can capture an ellipsoidal ham's geometry properties and thus facilitate the robotics vision for food products, especially in stimulating automatic industrial manufacturing development.

show abstract

Section: Proposed Frameworkmentioning

confidence: 93%

Find the centroid: A vision‐based approach for optimal object grasping

Liong

Tan

et al. 2021

J Food Process Engineering

View full text Add to dashboard Cite

show abstract

“…If we move to the effect of this on working with remote sensing data, specifically a land cover classification, it means a great deal of effort and money. [22]. Qiu et al have increased the data used in training by using different dates for these images in the same location, which means that we have an increasing number of spectral signatures for the same location due to the abundance of data available from the navigating satellites [27], They realized their idea using residual convolutional neural networks (ResNet).…”

Section: Related Workmentioning

confidence: 99%

“…Adadelta optimizer is one of the best optimizers used to train U-Net [22]. It is an extension to Adagrad algorithm as it reduces the extensive calculations by using a small and fixed window of past gradients.…”

Section:  Optimizer Functionmentioning

confidence: 99%

“…This ability helps us determine the exact location of each class in the image. Through the encoding (downsampling) path to obtain the class and the decoding (upsampling) path to obtain the location [22]. Therefore, U-Net is not a mere pixel-based method, but it takes into consideration the relationship within each area of the image through the consistency between neighbors in the training process.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Sparse Pixel Training of Convolutional Neural Networks for Land Cover Classification

et al. 2021

View full text Add to dashboard Cite

Convolutional Neural Networks (CNN) have become the core of modern machine learning approaches. In addition to its inspiring interior design idea, the success of CNN depends mainly on two factors, the first is the availability of training data and the second is the computing power of the used devices. In the field of remote sensing, data availability is difficult and expensive. Furthermore, processing large remote sensing data to accommodate different models is a laborious process. At the same time, training data is often collected in the form of points distributed over crop fields rather than regions, which results in the scarcity of training data. To specifically address the scarcity of training points, in this paper we present a sparse pixel-based training of U-Net convolutional neural networks for land cover classification. Training images are reconstructed from the points' collection in a random manner, they are used as an input for the convolution networks. Based on this proposed method, the amount of training data is reproduced from the different spectral signals for each land cover. We conducted extensive experiments on eight classes, using ground truth data collected from several locations in Fayoum Governorate, Egypt. The obtained results showed the superiority of the proposed method over other methods.

show abstract

“…Among the multiple deep neural networks, encoder-decoder architectures represented by U-Net [51], which are a so-called fully convolutional network, are supposed to mitigate challenge (2). It is computationally efficient and delivers good results with small amounts of ground-truth data [51][52][53], and therefore, is being increasingly employed to analyze high-resolution data collected using UAV [54][55][56][57][58].…”

Section: Introductionmentioning

confidence: 99%

Deep neural networks based automated extraction of dugong feeding trails from UAV images in the intertidal seagrass beds

et al. 2021

View full text Add to dashboard Cite

Dugongs (Dugong dugon) are seagrass specialists distributed in shallow coastal waters in tropical and subtropical seas. The area and distribution of the dugongs’ feeding trails, which are unvegetated winding tracks left after feeding, have been used as an indicator of their feeding ground utilization. However, current ground-based measurements of these trails require a large amount of time and effort. Here, we developed effective methods to observe the dugongs’ feeding trails using unmanned aerial vehicle (UAV) images (1) by extracting the dugong feeding trails using deep neural networks. Furthermore, we demonstrated two applications as follows; (2) extraction of the daily new feeding trails with deep neural networks and (3) estimation the direction of the feeding trails. We obtained aerial photographs from the intertidal seagrass bed at Talibong Island, Trang Province, Thailand. The F1 scores, which are a measure of binary classification model’s accuracy taking false positives and false negatives into account, for the method (1) were 89.5% and 87.7% for the images with ground sampling resolutions of 1 cm/pixel and 0.5 cm/pixel, respectively, while the F1 score for the method (2) was 61.9%. The F1 score for the method (1) was high enough to perform scientific studies on the dugong. However, the method (2) should be improved, and there remains a need for manual correction. The mean area of the extracted daily new feeding trails from September 12–27, 2019, was 187.8 m2 per day (n = 9). Total 63.9% of the feeding trails was estimated to have direction within a range of 112.5° and 157.5°. These proposed new methods will reduce the time and efforts required for future feeding trail observations and contribute to future assessments of the dugongs’ seagrass habitat use.

show abstract

U-Net Convolutional Networks for Mining Land Cover Classification Based on High-Resolution UAV Imagery

Cited by 60 publications

References 59 publications

Find the centroid: A vision‐based approach for optimal object grasping

Find the centroid: A vision‐based approach for optimal object grasping

Sparse Pixel Training of Convolutional Neural Networks for Land Cover Classification

Deep neural networks based automated extraction of dugong feeding trails from UAV images in the intertidal seagrass beds

Contact Info

Product

Resources

About