Detection Speed
Recently Published Documents





2021 ◽  
Vol 13 (21) ◽  
pp. 4213
Lingfei Shi ◽  
Feng Zhang ◽  
Junshi Xia ◽  
Jibo Xie ◽  
Zhe Zhang ◽  

The collapse of buildings caused by the earthquake seriously threatened human lives and safety. So, the quick detection of collapsed buildings from post-earthquake images is essential for disaster relief and disaster damage assessment. Compared with the traditional building extraction methods, the methods based on convolutional neural networks perform better because it can automatically extract high-dimensional abstract features from images. However, there are still many problems with deep learning in the extraction of collapsed buildings. For example, due to the complex scenes after the earthquake, the collapsed buildings are easily confused with the background, so it is difficult to fully use the multiple features extracted by collapsed buildings, which leads to time consumption and low accuracy of collapsed buildings extraction when training the model. In addition, model training is prone to overfitting, which reduces the performance of model migration. This paper proposes to use the improved classic version of the you only look once model (YOLOv4) to detect collapsed buildings from the post-earthquake aerial images. Specifically, the k-means algorithm is used to optimally select the number and size of anchors from the image. We replace the Resblock in CSPDarkNet53 in YOLOv4 with the ResNext block to improve the backbone’s ability and the performance of classification. Furthermore, to replace the loss function of YOLOv4 with the Focal-EOIU loss function. The result shows that compared with the original YOLOv4 model, our proposed method can extract collapsed buildings more accurately. The AP (average precision) increased from 88.23% to 93.76%. The detection speed reached 32.7 f/s. Our method not only improves the accuracy but also enhances the detection speed of the collapsed buildings. Moreover, providing a basis for the detection of large-scale collapsed buildings in the future.

2021 ◽  
Vol 13 (19) ◽  
pp. 3908
Zhenfang Qu ◽  
Fuzhen Zhu ◽  
Chengxiao Qi

Remote sensing image target detection is widely used for both civil and military purposes. However, two factors need to be considered for remote sensing image target detection: real-time and accuracy for detecting targets that occupy few pixels. Considering the two above issues, the main research objective of this paper is to improve the performance of the YOLO algorithm in remote sensing image target detection. The reason is that the YOLO models can guarantee both detection speed and accuracy. More specifically, the YOLOv3 model with an auxiliary network is further improved in this paper. Our model improvement consists of four main components. Firstly, an image blocking module is used to feed fixed size images to the YOLOv3 network; secondly, to speed up the training of YOLOv3, DIoU is used, which can speed up the convergence and increase the training speed; thirdly, the Convolutional Block Attention Module (CBAM) is used to connect the auxiliary network to the backbone network, making it easier for the network to notice specific features so that some key information is not easily lost during the training of the network; and finally, the adaptive feature fusion (ASFF) method is applied to our network model with the aim of improving the detection speed by reducing the inference overhead. The experiments on the DOTA dataset were conducted to validate the effectiveness of our model on the DOTA dataset. Our model can achieve satisfactory detection performance on remote sensing images, and our model performs significantly better than the unimproved YOLOv3 model with an auxiliary network. The experimental results show that the mAP of the optimised network model is 5.36% higher than that of the original YOLOv3 model with the auxiliary network, and the detection frame rate was also increased by 3.07 FPS.

Jiacheng Rong ◽  
Guanglin Dai ◽  
Pengbo Wang

AbstractFor automating the harvesting of bunches of tomatoes in a greenhouse, the end-effector needs to reach the exact cutting point and adaptively adjust the pose of peduncles. In this paper, a method is proposed for peduncle cutting point localization and pose estimation. Images captured in real time at a fixed long-distance are detected using the YOLOv4-Tiny detector with a precision of 92.7% and a detection speed of 0.0091 s per frame, then the YOLACT +  + Network with mAP of 73.1 and a time speed of 0.109 s per frame is used to segment the close-up distance. The segmented peduncle mask is fitted to the curve using least squares and three key points on the curve are found. Finally, a geometric model is established to estimate the pose of the peduncle with an average error of 4.98° in yaw angle and 4.75° in pitch angle over the 30 sets of tests.

Yuqi Pang ◽  
Gang Ma ◽  
Xiaotian Xu ◽  
Xunyu Liu ◽  
Xinyuan Zhang

Background: Fast and reliable fault detection methods are the main technical challenges faced by photovoltaic grid-connected systems through modular multilevel converters (MMC) during the development. Objective: Existing fault detection methods have many problems, such as the inability of non-linear elements to form accurate analytical expressions, the difficulty of setting protection thresholds and the long detection time. Method: Aiming at the problems above, this paper proposes a rapid fault detection method for photovoltaic grid-connected systems based on Recurrent Neural Network (RNN). Results: The phase-to-mode transformation is used to extract the fault feature quantity to get the RNN input data. The hidden layer unit of the RNN is trained through a large amount of simulation data, and the opening instruction is given to the DC circuit breaker. Conclusion: The simulation verification results show that the proposed fault detection method has the advantage of faster detection speed without difficulties in setting and complicated calculation.

2021 ◽  
Vol 11 (18) ◽  
pp. 8663
Wen Chen ◽  
Chengwei Ju ◽  
Yanzhou Li ◽  
Shanshan Hu ◽  
Xi Qiao

The rapid and accurate identification of sugarcane stem nodes in the complex natural environment is essential for the development of intelligent sugarcane harvesters. However, traditional sugarcane stem node recognition has been mainly based on image processing and recognition technology, where the recognition accuracy is low in a complex natural environment. In this paper, an object detection algorithm based on deep learning was proposed for sugarcane stem node recognition in a complex natural environment, and the robustness and generalisation ability of the algorithm were improved by the dataset expansion method to simulate different illumination conditions. The impact of the data expansion and lighting condition in different time periods on the results of sugarcane stem nodes detection was discussed, and the superiority of YOLO v4, which performed best in the experiment, was verified by comparing it with four different deep learning algorithms, namely Faster R-CNN, SSD300, RetinaNet and YOLO v3. The comparison results showed that the AP (average precision) of the sugarcane stem nodes detected by YOLO v4 was 95.17%, which was higher than that of the other four algorithms (78.87%, 88.98%, 90.88% and 92.69%, respectively). Meanwhile, the detection speed of the YOLO v4 method was 69 f/s and exceeded the requirement of a real-time detection speed of 30 f/s. The research shows that it is a feasible method for real-time detection of sugarcane stem nodes in a complex natural environment. This research provides visual technical support for the development of intelligent sugarcane harvesters.

Agriculture ◽  
2021 ◽  
Vol 11 (9) ◽  
pp. 863
Chenglong Wang ◽  
Zhifeng Xiao

Food defect detection is crucial for the automation of food production and processing. Potato surface defect detection remains challenging due to the irregular shape of potato individuals and various types of defects. This paper employs deep convolutional neural network (DCNN) models for potato surface defect detection. In particular, we applied transfer learning by fine-tuning a base model through three DCNN models—SSD Inception V2, RFCN ResNet101, and Faster RCNN ResNet101—on a self-developed dataset, and achieved an accuracy of 92.5%, 95.6%, and 98.7%, respectively. RFCN ResNet101 presented the best overall performance in detection speed and accuracy. It was selected as the final model for out-of-sample testing, further demonstrating the model’s ability to generalize.

2021 ◽  
Vol 23 (08) ◽  
pp. 457-461
Sudhakar K ◽  
Dr.Subhash Kulkarni ◽  

This paper presents the performance evaluation of various distance metric in copy move forger detection algorithms. The choice of distance metric affects the detection speed. The proposed approach is tested over 9 different distance metrics. The experimental results found indicate the choice of distance metric has a considerable impact on forgery detection speed.

2021 ◽  
Lu Tan ◽  
Tianran Huangfu ◽  
Liyao Wu ◽  
Wenying Chen

Abstract Background: The correct identification of pills is very important to ensure the safe administration of drugs to patients. We used three currently mainstream object detection models, respectively Faster R-CNN, Single Shot Multi-Box Detector (SSD), and You Only Look Once v3(YOLO v3), to identify pills and compare the associated performance.Methods: In this paper, we introduce the basic principles of three object detection models. We trained each algorithm on a pill image dataset and analyzed the performance of the three models to determine the best pill recognition model. Finally, these models are then used to detect difficult samples and compare the results.Results: The mean average precision (MAP) of Faster R-CNN reached 87.69% but YOLO v3 had a significant advantage in detection speed where the frames per second (FPS) was more than eight times than that of Faster R-CNN. This means that YOLO v3 can operate in real time with a high MAP of 80.17%. The YOLO v3 algorithm also performed better in the comparison of difficult sample detection results. In contrast, SSD did not achieve the highest score in terms of MAP or FPS.Conclusion: Our study shows that YOLO v3 has advantages in detection speed while maintaining certain MAP and thus can be applied for real-time pill identification in a hospital pharmacy environment.

Shang Jiang ◽  
Haoran Qin ◽  
Bingli Zhang ◽  
Jieyu Zheng

The loss function is a crucial factor that affects the detection precision in the object detection task. In this paper, we optimize both two loss functions for classification and localization simultaneously. Firstly, we reconstruct the classification loss function by combining the prediction results of localization, aiming to establish the correlation between localization and classification subnetworks. Compared to the existing studies, in which the correlation is only established among the positive samples and applied to improve the localization accuracy of predicted boxes, this paper utilizes the correlation to define the hard negative samples and then puts emphasis on the classification of them. Thus the whole misclassified rate for negative samples can be reduced. Besides, a novel localization loss named MIoU is proposed by incorporating a Mahalanobis distance between the predicted box and target box, eliminating the gradients inconsistency problem in the DIoU loss, further improving the localization accuracy. Finally, the proposed methods are applied to train the networks for nighttime vehicle detection. Experimental results show that the detection accuracy can be outstandingly improved with our proposed loss functions without hurting the detection speed.

Sign in / Sign up

Export Citation Format

Share Document