Search

Abstract An increasing number of researchers are using deep learning technology to classify and process garbage in rural areas, and have achieved certain results. However, the existing garbage detection models still have problems such as high complexity, missed detection of small targets, low detection accuracy and poor real-time performance. To address these issues, we train a model and apply it to garbage classification and detection in rural areas. In general, we propose an attention combination mechanism based on the YOLOv5 algorithm to build a better backbone network structure, add a new small object detection layer in the head network to enhance the model's ability to detect small objects, adopt the CIoU loss function to optimize the output prediction bounding box, and choose the Adam optimization algorithm to train the model. Our proposed YOLOv5s-CSS model detects a single garbage image in 0.021 s with a detection accuracy of 96.4%. Compared with the YOLOv5 algorithm and the classic detection algorithm, the improved algorithm has better detection speed and detection accuracy. At the same time, the complexity of the network model is reduced to a certain extent, which can meet the requirements of real-time detection of rural domestic garbage.

It is well-known that geomagnetic fields have multiple components or parameters, and that these geomagnetic parameters are related to each other. In this paper, a parameter selection method is proposed, and this paper mainly discusses the correlation of geomagnetic field parameters for geomagnetic navigation technology. For the correlation analysis between geomagnetic parameters, the similarity calculation of the correlation coefficient is firstly introduced for geomagnetic navigation technology, and the grouped results are obtained by data analysis. At the same time, the search algorithm (Hex-path algorithm) is used to verify the correlation analysis results. The results show the same convergent state for the approximate correlation coefficient. In other words, the simulation results are in agreement with the similarity calculation results.

An aggregation or assembly of Ag triangular nanoplates (Ag TNPs) can cause dramatic changes in their optical properties, which is widely used in applications in the field of sensing. The assembly forms of nanoparticles are crucial for obtaining sensitive sensing signals, but it is unknown what kind of assembly dominates the aggregated Ag TNPs in aqueous solutions. Herein, using thiram-induced Ag TNP aggregation as a model, six different assembly models were established, including three planar (side-by-side, side-to-tip, and tip-to-tip) assemblies and three tridimensional (plane-to-plane, plane-to-tip, and plane-to-side) assemblies. The corresponding optical properties were then investigated. Both theoretical and experimental findings indicate that three-dimensional assemblies, especially plane-to-plane assembly, dominate the Ag TNPs aggregation solution, causing a blue shift of the absorption spectrum. Analysis of charge distribution patterns in Ag TNPs indicates that such a blue shift is caused by the electrostatic repulsive force in plane-to-plane assembly. Thus, we propose a simple colorimetric method for thiram detection using Ag TNPs as an indicator. The method exhibits a selective and sensitive response to thiram with a limit of detection of 0.13 μM in the range of 0.2–0.5 μM, as well as excellent performance in real samples like wheat.

A novel fuzzy control scheme with adaptation algorithms is developed for robot manipulators’ system. At the beginning, one adjustable parameter is introduced in the fuzzy logic system, the robot manipulators system with uncertain nonlinear terms as the master device and a reference model dynamic system as the slave robot system. To overcome the limitations such as online learning computation burden and logic structure in conventional fuzzy logic systems, a parameter should be used in fuzzy logic system, which composes fuzzy logic system with updated parameter laws, and can be formed for a new fashioned adaptation algorithms controller. The error closed-loop dynamical system can be stabilized based on Lyapunov analysis, the number of online learning computation burdens can be reduced greatly, and the different kinds of fuzzy logic systems with fuzzy rules or without any fuzzy rules are also suited. Finally, effectiveness of the proposed approach has been shown in simulation example.

To address the challenges of high complexity, poor real-time performance, and low detection rates for small target vehicles in existing vehicle object detection algorithms, this paper proposes a real-time lightweight architecture based on You Only Look Once (YOLO) v5m. Firstly, a lightweight upsampling operator called Content-Aware Reassembly of Features (CARAFE) is introduced in the feature fusion layer of the network to maximize the extraction of deep-level features for small target vehicles, reducing the missed detection rate and false detection rate. Secondly, a new prediction layer for tiny targets is added, and the feature fusion network is redesigned to enhance the detection capability for small targets. Finally, this paper applies L1 regularization to train the improved network, followed by pruning and fine-tuning operations to remove redundant channels, reducing computational and parameter complexity and enhancing the detection efficiency of the network. Training is conducted on the VisDrone2019-DET dataset. The experimental results show that the proposed algorithmreduces parameters and computation by 63.8% and 65.8%, respectively. The average detection accuracy improves by 5.15%, and the detection speed reaches 47 images per second, satisfying real-time requirements. Compared with existing approaches, including YOLOv5m and classical vehicle detection algorithms, our method achieves higher accuracy and faster speed for real-time detection of small target vehicles in edge computing. [ABSTRACT FROM AUTHOR]

To address the issue of imbalanced detection performance and detection speed in current mainstream object detection algorithms for optical remote sensing images, this paper proposes a multi-scale object detection model for remote sensing images on complex backgrounds, called DI-YOLO, based on You Only Look Once v7-tiny (YOLOv7-tiny). Firstly, to enhance themodel’s ability to capture irregular-shaped objects and deformation features, as well as to extract high-level semantic information, deformable convolutions are used to replace standard convolutions in the original model. Secondly, a Content Coordination Attention Feature Pyramid Network (CCAFPN) structure is designed to replace the Neck part of the original model, which can further perceive relationships between different pixels, reduce feature loss in remote sensing images, and improve the overall model’s ability to detect multi-scale objects. Thirdly, an Implicitly Efficient Decoupled Head (IEDH) is proposed to increase the model’s flexibility, making it more adaptable to complex detection tasks in various scenarios. Finally, the Smoothed Intersection over Union (SIoU) loss function replaces the Complete Intersection over Union (CIoU) loss function in the original model, resulting in more accurate prediction of bounding boxes and continuous model optimization. Experimental results on the High-Resolution Remote Sensing Detection (HRRSD) dataset demonstrate that the proposed DI-YOLO model outperforms mainstream target detection algorithms in terms of mean Average Precision (mAP) for optical remote sensing image detection. Furthermore, it achieves Frames Per Second (FPS) of 138.9, meeting fast and accurate detection requirements. [ABSTRACT FROM AUTHOR]

A forest is an ensemble with decision trees as members. This paper proposes a novel strategy to pruning forest to enhance ensemble generalization ability and reduce ensemble size. Unlike conventional ensemble pruning approaches, the proposed method tries to evaluate the importance of branches of trees with respect to the whole ensemble using a novel proposed metric called importance gain. The importance of a branch is designed by considering ensemble accuracy and the diversity of ensemble members, and thus the metric reasonably evaluates how much improvement of the ensemble accuracy can be achieved when a branch is pruned. Our experiments show that the proposed method can significantly reduce ensemble size and improve ensemble accuracy, no matter whether ensembles are constructed by a certain algorithm such as bagging or obtained by an ensemble selection algorithm, no matter whether each decision tree is pruned or unpruned. [ABSTRACT FROM AUTHOR]