Locating the tiny insulator defect object with complex backgrounds in high-resolution aerial images is a challenging task. In this paper, we propose a novel method which cascades detection and segmentation networks to identify the defect from the global and local two levels: (1) The improved Faster R-CNN is carried out to capture both defects and insulators in the entire image. ResNeXt-101 is adopted as the feature extraction network so as to fully extract features, and Feature Pyramid Network (FPN) is built to enhance the ability of detecting small targets. In addition, the Online Hard Example Mining (OHEM) training strategy is applied to solve the imbalance problem of positive and negative samples. (2) All the detected insulators are extracted and fed into the improved U-Net network to futher inspect at pixel level, we utilize the pre-trained ResNeXt-50 as the encoder of U-Net, incorporate an attention module, Spatial and Channel Squeeze & Excitation Block (SCSE), into the decoding path to highlight the meaningful information. A hybrid loss which merges binary cross entropy (BCE) loss and dice coefficient loss is designed to train our network for figuring out the class imbalance issue. The missed detection can be greatly reduced with the combination of two modified network, which makes comprehensive use of the original map information and local information. On the test set of actual images, the insulator defect recognition precision and recall of the cascade network is 91.9% and 95.7%, exhibiting strong robustness and accuracy.

No abstract

Summary Parallel computing is an effective method to solve computationally large and data‐intensive problems. The traditional data mining algorithm cannot mining association rules for large amounts of streaming data in a timely and effectively. In order to improve the speed and accuracy of association rules mining, distributed and parallel algorithms have become a research focus. This paper proposes a parallel FP‐growth approach using Spark Streaming, called SSPFP, which can parallel mining frequent itemsets and association rules in real‐time streaming data. In this paper, the proposed SSPFP algorithm is applied to mining the association rules between temperature and salinity in marine Argo data. The experimental results indicate that SSPFP algorithm is efficient for association rules mining.

No abstract

With the increase of vehicles and the diversification of road conditions, people pay more attention to the safety of driving. In recent years, autonomous driving technology by Franke et al. (IEEE Intell Syst Their Appl 13(6):40-48, 1998) and unmanned driving technology by Zhang et al. (CAAI Trans Intell Technol 1(1):4-13, 2016) have entered our field of vision. Both automatic driving by Levinson et al. (Towards fully autonomous driving: Systems and algorithms, 2011) and unmanned driving by Im et al. (Unmanned driving of intelligent robotic vehicle, 2009) use a variety of sensors to collect the environment around the vehicle, and use a variety of decision control algorithms to control the vehicle in motion. The visual driving assistance system by Watanabe, et al. (Driving assistancesystem for appropriately making the driver recognize another vehicle behind or next to present vehicle, 2010), used in conjunction with the target recognition algorithm by Pantofaru et al. (Object recognition by integrating multiple image segmentations, 2008)), will provide drivers with real-time environment around the vehicle. In recent years, few-shot learning by Li et al. (Comput Electron Agric 2:2, 2020) has become a new direction of target recognition algorithm, which reduces the difficulty of collecting training samples.In this paper, on one hand, several low-light cameras with fish-eye lenses are used to collect and reconstruct the environment around the vehicle. On the other hand, we use infrared camera and lidar to collect the environment in front of the vehicle. Then, we use the method of few-shot learning to identify vehicles and pedestrians in the forward-view image. In addition, we develop the system on embedded devices according to miniaturization requirements. In conclusion, the system will adapt to the needs of most drivers at this stage, and will effectively cooperate with the development of automatic driving and unmanned driving.