Abstract：Crude oil, as an important strategic material, plays an important role in many fields such as my country′s economy and military. This paper proposes an algorithm MSA-YOLO(MultiScale Adaptive YOLO), which is optimized on the basis of the YOLOv4 algorithm, and is experimented based on the remote sensing image dataset mainly based on Jilin-1 optical remote sensing satellite images，to make identification and classification of oil storage tanks. The algorithm optimization contents include: in order to simplify the oil storage tank monitoring model and ensure the efficiency of the model, prune the multi-scale identification module in the network structure of YOLOv4; use the k-means++ clustering algorithm to select the initial anchor frame to accelerate the convergence of the model;use CIoU-NMS-based optimization to further improve inference speed and accuracy. The experimental results show that compared with YOLOv4, the number of parameters of MSA-YOLO model is reduced by 25.84%; the model size is reduced by 62.13%; in the GPU environment of Tesla V100, the training speed of the model is increased by 6 s/epoch, and the inference speed is increased by 15.76 F/s; the average accuracy is 95.65%. At the same time, the MSA-YOLO algorithm shows more efficient characteristics in the comparative experiments of various general target recognition algorithms. The MSA-YOLO algorithm has universal feasibility for accurate and real-time identification of oil storage tanks, and can provide technical reference for remote sensing data in the field of energy futures.

Abstract：Aiming at the problems of false positive detection and low recall in the detection of small targets by the general target detection algorithm, a small target detection algorithm IPH(involution prediction head) is proposed, which is applied to the detection head of YOLOv4 and YOLOv5. The experimental results on the VOC2007 data set show that the detection accuracy APs(AP for small objects) of YOLOv4 after using IPH is improved by 1.1% compared with the original algorithm, and the APs on YOLOv5 is improved by 5.9%. Through further verification of the intelligent traffic detection data set, IPH algorithm and desampling can effectively improve the accuracy of small object detection and reduce false positive detection and missed detection.

Abstract：Aiming at the problem of poor recognition accuracy of license plate recognition system in haze scene, an improved license plate recognition model is proposed. The model uses the improved dark channel apriori defogging algorithm for defogging. Considering the color distortion and other problems when the original defogging algorithm processes the haze image with bright areas, firstly, the atmospheric light value is limited by the threshold value. Secondly, the introduction factor is optimized. And finally, the tolerance mechanism is introduced to correct the transmittance, and the image brightness is adjusted to improve the image visualization effect. The simulation results show that the performance of PSNR, SSIM, enterprise and e improved by 1.934 dB, 0.082, 0.235 and 38.995 respectively. The recognition test of the license plate image before and after defogging shows that the recognition accuracy of the license plate is improved by 22%, which proves the superiority of the proposed model.

Abstract：In order to improve the screening and diagnosis efficiency of hepatic hydatid disease, and make up for the shortage of medical resources in some areas, this paper proposes an intelligent typing method of hepatic hydatid disease based on Swin Transformer, which combines the convolution attention mechanism model, and realizes the automatic classification of five types of cystic hydatid disease by learning the whole and local details of images. In order to verify the superiority of the model, the prediction model proposed in this paper is compared with common classification models. The results show that the classification accuracy based on the improved Swin Transformer model can reach 92.6% on the test set. The experimental results show that compared with other algorithms, the improved Swin Transformer network can better classify the ultrasonic images of hepatic cystic echinococcosis, and this method can be extended to other medical applications.

Abstract：Seal recognition is an indispensable part of the intelligent office. The current stage of seal recognition method is to directly input the scanned electronic documents into the neural network model for identification, facing the problems of unable to accurately locate the position of the seal, low accuracy of bending text recognition. Aiming at the above problems, this paper proposes an efficient stamp text recognition method, which uses the diffuse water filling algorithm to process the grayscale image for seal image feature enhancement, which ensures the accuracy of Chinese seal detection, and introduces the polar coordinate conversion operation to ensure the integrity of text features. In order to evaluate the effectiveness of the proposed method, multiple sets of comparative experiments were carried out in the existing and other network models. Experimental results show that the existing network model fused the text features extracted by the method shows excellent recognition results.