Accurate and efficient real-time detection of orchard pests was essential and could improve the economic benefits of the fruit industry. The orchard pest dataset, PestImgData, was built through a series of methods such as web crawler, specimen image collection and data augmentation. PestImgData was composed of two parts, PestImgData-1 and PestImgData-2. It contained 24,796 color images and covered 7 types of orchard pests. Based on the PestImgData and YOLOv4 algorithm, this paper conducted a preliminary study on the real-time object detection of orchard pests from 4 perspectives: transfer learning, activation function, anchor box, and batch normalization. In addition, this paper also visualized the feature learning ability of the detection models. On the basis of the above research, three improvement measures were adopted: the post-processing NMS algorithm was upgraded to DIoU-NMS, the training method was upgraded to 2-time finetuning training and the training data was enhanced. The performance of the improved model, F-D-YOLOv4-PEST, had been effectively improved. The mean average precision of F-D-YOLOv4-PEST was 92.86%, and the detection time of a single picture was 12.22 ms, which could meet the real-time detection requirements. In addition, in the case of high overlap area or high density, F-D-YOLOv4-PEST still maintained good performance. In the testing process of the laboratory and the greenhouse, including the wired network and the wireless network, F-D-YOLOv4-PEST could locate and classify pests as expected. This research could provide technical reference for the intelligent identification of agricultural pests based on deep learning.

Motor imagery brain‐computer interfaces (BCIs) have demonstrated great potential and attract world‐spread attentions. Due to the nonstationary character of the motor imagery signals, costly and boring calibration sessions must be proceeded before use. This prevents them from going into our realistic life. In this paper, the source subject’s data are explored to perform calibration for target subjects. Model trained on source subjects is transferred to work for target subjects, in which the critical problem to handle is the distribution shift. It is found that the performance of classification would be bad when only the marginal distributions of source and target are made closer, since the discriminative directions of the source and target domains may still be much different. In order to solve the problem, our idea comes that joint distribution adaptation is indispensable. It makes the classifier trained in the source domain perform well in the target domain. Specifically, a measure for joint distribution discrepancy (JDD) between the source and target is proposed. Experiments demonstrate that it can align source and target data according to the class they belong to. It has a direct relationship with classification accuracy and works well for transferring. Secondly, a deep neural network with joint distribution matching for zero‐training motor imagery BCI is proposed. It explores both marginal and joint distribution adaptation to alleviate distribution discrepancy across subjects and obtain effective and generalized features in an aligned common space. Visualizations of intermediate layers illustrate how and why the network works well. Experiments on the two datasets prove the effectiveness and strength compared to outstanding counterparts.

Real-time detection and identification of orchard pests is related to the economy of the orchard industry. Using lab picture collections and pictures from web crawling, a dataset of common pests in orchards has been created. It contains 24,748 color images and covers seven types of orchard pests. Based on this dataset, this paper combines YOLOv5 and GhostNet and explains the benefits of this method using feature maps, heatmaps and loss curve. The results show that the mAP of the proposed method increases by 1.5% compared to the original YOLOv5, with 2$\times$ or 3$\times$ fewer parameters, less GFLOPs and the same or less detection time. Considering the fewer parameters of the Ghost convolution, our new method can reach a higher mAP with the same epochs. Smaller neural networks are more feasible to deploy on FPGAs and other embedding devices which have limited memory. This research provides a method to deploy the algorithm on embedding devices.

QR code payment plays an indispensable role in the mobile payment market, and the security of scanning codes has always been a problem in the field of information security. Static QR codes are easily copied and replaced, and there are huge security loopholes. The QR code payment in a closed system still faces security challenges. In order to solve the security problem of QR code payment, we have studied dynamic QR code payment system that supports SM2, SM3, and SM4 cryptographic algorithms, which can realize QR code scanning and scanned transactions, UnionPay cloud QuickPass transactions, etc., and generate dynamic QR code information in real time during the transaction process, one order and one code. Through dynamic algorithm distribution, the randomness and uniqueness of QR code generation are guaranteed, and it is suitable for multi-scene application transactions. The algorithm correctness test result shows that the system has achieved the expected effect. The performance test results show that the hardware of the security module implements the algorithm flow and improves the payment performance. Compared with some other algorithms, the processing time is shorter, the running speed is faster, and the system is more secure.

Nowadays, major depressive disorder (MDD) has become a crucial mental disease that endangers human health. Good results have been achieved by electroencephalogram (EEG) signals in the detection of depression. However, EEG signals are time-varying, and the distributions of the different subjects’ data are non-uniform, which poses a bad influence on depression detection. In this paper, the deep learning method with domain adaptation is applied to detect depression based on EEG signals. Firstly, the EEG signals are preprocessed and then transformed into pictures by two methods: the first one is to present the three channels of EEG separately in the same image, and the second one is the RGB synthesis of the three channels of EEG. Finally, the training and prediction are performed in the domain adaptation model. The results indicate that the domain adaptation model can effectively extract EEG features and obtain an average accuracy of 77.0 ± 9.7%. This paper proves that the domain adaptation method can effectively weaken the inherent differences of EEG signals, making the diagnosis of different users more accurate.

High-sensitivity detection of vibrations under high temperatures is a topic of great interest in modern engineering such as thermal engine deep-sea aquaculture factory ship, aerospace, high temperature casting, energy, etc. As traditional accelerometers and some fiber optic F-P accelerometers have shown their sensing limits at about 400 °C and 650 °C, respectively, a high temperature fiber optic F-P accelerometer based on MEMS technology is proposed. To obtain a high-performance chip for the sensor, an examination of the theoretical performance of an L and Г-shaped cantilever beam diaphragm shows a sensitivity of 15.05 nm/g and 53.7 nm/g, respectively, and a wide working frequency range. Thanks to the designed sensor’s various protections, frequency measurements with a high-temperature performance of 850 °C are recorded. The L-shaped cantilever beams diaphragm allows the sensor measurements at 850 °C with a repeatability of 5.46%, a working frequency range of 100-1000 Hz, an experimental sensitivity of 389 mV/g, an overall stability of 8 jumps at its adjacent frequency resolution range over 150 measurements, a linearity of 0.9856 and a maximum relative error maintained below 1.72%. In the field of application, it also exhibits a good relative error of measurement respecting the technical specification of 5 Hz.