Obstacle type recognition in visual images via dilated convolutional neural network for unmanned surface vehicles

Shi, Binghua; Lian, Cheng; Xiong, Chang; Long, Yang; Gong, Chenglong

doi:10.1017/s0373463321000941

Cited by 7 publications

(7 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To enrich the diversity of the testing scenarios, the effectiveness of the proposed method is verified on the classical Singapore marine dataset (SMD), MODD, and our self-collected Yangtze River navigation scene dataset (YRNSD) 36 . The classical SMD contains navigation scenes in various environmental conditions obtained using a Canon 70D camera in waters around Singapore from July 2015 to May 2016 13 .…”

Section: Experiments and Resultsmentioning

confidence: 99%

Research on the visual image-based complexity perception method of autonomous navigation scenes for unmanned surface vehicles

Shi

Guo

Wang³

et al. 2022

Sci Rep

Self Cite

View full text Add to dashboard Cite

To solve the long-tail problem and improve the testing efficiency for autonomous navigation systems of unmanned surface vehicles (USVs), a visual image-based navigation scene complexity perception method is proposed. In this paper, we intend to accurately construct a mathematical model between navigation scene complexity and visual features from the analysis and processing of image textures. First, the typical complex elements are summarized, and the navigation scenes are divided into four levels according to whether they contain these typical elements. Second, the textural features are extracted using the gray level cogeneration matrix (GLCM) and Tamura coarseness, which are applied to construct the feature vectors of the navigation scenes. Furthermore, a novel paired bare bone particle swarm clustering (PBBPSC) method is proposed to classify the levels of complexity, and the exact value of the navigation scene complexity is calculated using the clustering result and an interval mapping method. By comparing different methods on the classical and self-collected datasets, the experimental results show that our proposed complexity perception method can not only better describe the level of complexity of navigation scenes but also obtain more accurate complexity values.

show abstract

Section: Experiments and Resultsmentioning

confidence: 99%

Research on the visual image-based complexity perception method of autonomous navigation scenes for unmanned surface vehicles

Shi

Guo

Wang³

et al. 2022

Sci Rep

Self Cite

View full text Add to dashboard Cite

show abstract

“…To verify the effectiveness of the proposed method, experiments were conducted on the SMD [18], MODD [19], and YRNSD datasets [6], of which the first two are classical datasets and the third is a self-collected dataset. The SMD dataset was collected in Singapore waters from July 2015 to May 2016 under various environmental conditions, such as before sunrise (40 min before sunrise), sunrise, midday, afternoon, evening, haze and rainfall, and after sunset (2 h after sunset) .…”

Section: Datasets and Evaluation Metricmentioning

confidence: 99%

“…Compared to laser rangefinder instruments, synthetic aperture radar vision sensors have extensive advantages, such as data richness, low cost, and good stability [2][3][4]. A USV with autonomous navigation capabilities can ensure self-safe and efficient execution to complete specific tasks; in particular, the rapid development of visual navigation technology has laid an important foundation for autonomous navigation [5,6]. For long voyages, early determination of the sea level relies on visual navigation technology to help maintain USV balance and smooth navigation [7,8].…”

Section: Introductionmentioning

confidence: 99%

Research on a Horizon Line Detection Method for Unmanned Surface Vehicles in Complex Environments

Shi

Guo

et al. 2023

JMSE

Self Cite

View full text Add to dashboard Cite

A critical step in the visual navigation of unmanned surface vehicles (USVs) is horizon line detection, which can be used to adjust the altitude as well as for obstacle avoidance in complex environments. In this paper, a real-time and accurate detection method for the horizon line is proposed. Our approach first differentiates the complexity of navigational scenes using the angular second moment (ASM) parameters in the grey level co-occurrence matrix (GLCM). Then, the region of interest (ROI) is initially extracted using minimal human interaction for the complex navigation scenes, while subsequent frames are dynamically acquired using automatic feature point matching. The matched ROI can be maximally removed from the complex background, and the Zernike-moment-based edges are extracted from the obtained ROI. Finally, complete sea horizon information is obtained through a linear fitting of the lower edge points to the edge information. Through various experiments carried out on a classical dataset, our own datasets, and that of another previously published paper, we illustrate the significance and accuracy of this technique for various complex environments. The results show that the performance has potential applications for the autonomous navigation and control of USVs.

show abstract

“…The combination of ResNet and U-Net can improve the accuracy of image detection and network training [19]. Dilated convolution can enlarge the receptive led and improve the target detection ability of the network [20].…”

Section: Introductionmentioning

confidence: 99%

Crop Insect Pest Detection based on Dilated Multi-scale Attention U-Net

Wang,

Zhang,

Zhang

2024

Preprint

View full text Add to dashboard Cite

Early detection and identification of insect pests is the premise and basis of scientific control and accurate utilization of Insect pesticides. Aiming at the problems of low detection accuracy and slow training speed of the existing crop Insect pest detection models, a dilated multi-scale attention U-Net (DMSAU-Net) model is constructed for crop Insect pest detection. In its encoder, dilated Inception is designed to replace the convolution layer in U-Net to extract the multi-scale features of insect pest images and improve the accuracy of the model. An attention module is added to its decoder to focus on the edge of the insect pest image and reduce the upsampling noise and accelerate model convergence. The results on the crop insect pest image dataset verify that the proposed method has high segmentation accuracy and good generalization ability, and can be applied to practical crop insect pest monitoring system.

show abstract

Obstacle type recognition in visual images via dilated convolutional neural network for unmanned surface vehicles

Cited by 7 publications

References 30 publications

Research on the visual image-based complexity perception method of autonomous navigation scenes for unmanned surface vehicles

Research on the visual image-based complexity perception method of autonomous navigation scenes for unmanned surface vehicles

Research on a Horizon Line Detection Method for Unmanned Surface Vehicles in Complex Environments

Crop Insect Pest Detection based on Dilated Multi-scale Attention U-Net

Contact Info

Product

Resources

About