Stagnant water on roads has always been a major cause of traffic jams and accidents. Traditional urban waterlogging monitoring and warning system is mainly based on a large amount of historical data and predictive network, which has low accuracy and weak generalization ability. Considering the deep neural network algorithms have demonstrated strong capabilities in computer vision tasks such as object detection, we aim to apply them to road stagnant water detection. In this paper, a novel automatic stagnant water localization method under weak supervision based on visual image is proposed. First, the template matching method is applied to extract road information from the traffic image. Then, due to the complexity of data annotation, we locate stagnant water in image based on Class Activation Maps (CAM) mechanism, which is a weakly supervised method. The detection model consists of the ResNet-18 and the Grad-CAM++ mechanism. Finally, based on the heat map and template, we set a suitable threshold to segment stagnant water area in image. In the experiments, the precision and recall for road stagnant water classification by the proposed model are 99.39% and 99.60%, while the Intersection over Union (IoU) for stagnant water area segmentation is up to 63%. These show that our method is effective for road stagnant water localization.
Computational Visual Media - Visual curve completion is a fundamental problem in understanding the principles of the human visual system. This problem is usually divided into two problems: a... 相似文献
Metallurgical and Materials Transactions B - In the past two decades, friction stir processing (FSP) technology has received considerable attention. FSP can be used to adjust and control the... 相似文献
The corrosion resistance, interfacial contact resistance (ICR), and hydrophobicity of cathodic arc deposited TiN films on 316L stainless steels at different nitrogen flow rates as bipolar plates (BPs) for the proton-exchange membrane fuel cell (PEMFC) are investigated. It is shown in the results that the TiN-coated 316L stainless steel prepared at a moderate nitrogen flow rate (250 sccm) exhibits a low stable corrosion current density of 6.5 × 10−8 A cm−2 in the simulated corrosive cathode environment of PEMFC, and a low ICR of 8.94 mΩ cm−2 at a pressure of 1.38 MPa after corrosion, meanwhile presents a good hydrophobicity before and after corrosion. These results are discussed by considering the probable effects of the nitrogen flow rate on the substrate/coating system based on the microstructural characterization of the substrate/coating interface and the coating, which shows that the interdiffusion will be started in the deposition process and a moderate nitrogen flow rate during the coating process will promote to the broadening of interface region and lead to the formation of a robust and high-quality coating with fewer defects that can effectively improve the performances of the 316L stainless steel substrate as the BP for PEMFC. 相似文献
本文提出了一种对分布式光纤声传感器的入侵事件分类方法.该方式采用小波包去噪方式对原始信号进行去噪;将去噪后的原始信号进行小波变换,得到原始信号的小波时频图;构建双输入型的卷积神经网络,将滤波后的原始一维信号直接输入到一个三层的1-D CNN中、滤波后得到的二维小波时频图直接输入到一个两层的2-D CNN中;将两种CNN输出的特征输入到支持向量机(SVM),使用SVM对事件进行分类.本文中主要识别3种振动事件:汽车通过、挖掘机挖掘和破路机工作.实验结果表明,所提方式对实际环境中3种振动事件的识别准确率平均可以达到96％,并且识别时间仅为0.61 s. 相似文献
