MPMR: Multi-Scale Feature and Probability Map for Melanoma Recognition

Malignant melanoma (MM) recognition in whole-slide images (WSIs) is challenging due to the huge image size of billions of pixels and complex visual characteristics. We propose a novel automatic melanoma recognition method based on the multi-scale features and probability map, named MPMR. First, we introduce the idea of breaking up the WSI into patches to overcome the difficult-to-calculate problem of WSIs with huge sizes. Second, to obtain and visualize the recognition result of MM tissues in WSIs, a probability mapping method is proposed to generate the mask based on predicted categories, confidence probabilities, and location information of patches. Third, considering that the pathological features related to melanoma are at different scales, such as tissue, cell, and nucleus, and to enhance the representation of multi-scale features is important for melanoma recognition, we construct a multi-scale feature fusion architecture by additional branch paths and shortcut connections, which extracts the enriched lesion features from low-level features containing more detail information and high-level features containing more semantic information. Fourth, to improve the extraction feature of the irregular-shaped lesion and focus on essential features, we reconstructed the residual blocks by a deformable convolution and channel attention mechanism, which further reduces information redundancy and noisy features. The experimental results demonstrate that the proposed method outperforms the compared algorithms, and it has a potential for practical applications in clinical diagnosis.

Application of an artificial neural network for determination of the cylinder-piston group wear level of the automobile-tractor engine

The article proposes reducing the redundancy of the neural network and the need to reduce the number of neurons in the hidden layer for a given level of network learning error. The minimum number of neurons of the hidden layer for the case of 11 monitoring standard sensors, the parameters of the automobile and tractor engine (ATE) and five classes of typical defects of the ATE nodes can be reduced to 5-7 with a high quality of recognition of the state of the ATE engine. The goal is to provide an expanded reliable knowledge base, the speed of information processing, the accuracy of the resulting technical diagnosis and the ability to quickly determine the technical state of an automotive engine in the mode real time. The basis of the proposed method is to ensure obtaining an extended reliable knowledge base, the speed of information processing, the accuracy of the obtained technical diagnosis and the ability to quickly determine the technical state of an ATE engine in real time. A feature of the proposed method is the use of voltages obtained in an artificial neural network from sensors that are standard in an ATE engine as input signals, and additionally indicate the output signal of the fuel cut-off device, provided for one step, containing a winding of a normally closed electromagnetic valve, which redirects fuel to the drain line. The use of the algorithm for identifying the values of the indicators of operating modes and malfunctions of the cylinder-piston group is the result of the analysis of an artificial neural network, which receive the results of the diagnostic parameters of the automotive engine. Having studied the artificial neural network 1 with different volumes of training data, we obtained the dependence of the change in the reliability of the result on the size of the training data and the reliability of the recognition result is 91.2%, the optimal amount of training data is 1200. Having examined the artificial neural network 2 with different volumes of training data, we obtained the dependence of the change in the reliability the result from the size of the training data and the reliability of the recognition result is 86.5%, the optimal amount of training data is from 10 to 15. The results obtained show the fundamental possibility of creating predictive models of units and assemblies of the tested automotive engines. The model can be created using the apparatus of artificial neural networks and using a fairly large database of tests performed.

ALGORITHM FOR PRECISE LANDING OF MULTIROTOR UNMANNED AIRCRAFT SYSTEMS AT AN AUTONOMOUS CHARGING STATION

In this article, we propose an algorithm for accurately landing multirotor (quadcopters, hexacopters, etc.) unmanned aerial vehicles (UAVs) at an autonomous charging station. This article also presents methods for locating the charging station and landing the UAV at night. Section 1 describes the general sequential landing procedures. Section 2 describes methods for detecting the ArUco marker and evaluating its position and orientation using the OpenCV computer vision library and shows the recognition result. In section 3, the precise landing algorithm is analyzed in detail, and a block diagram of the algorithm is given. Section 4 discusses the integration of the night vision camera into the landing algorithm.

Image-Based Hot Pepper Disease and Pest Diagnosis Using Transfer Learning and Fine-Tuning

Past studies of plant disease and pest recognition used classification methods that presented a singular recognition result to the user. Unfortunately, incorrect recognition results may be output, which may lead to further crop damage. To address this issue, there is a need for a system that suggest several candidate results and allow the user to make the final decision. In this study, we propose a method for diagnosing plant diseases and identifying pests using deep features based on transfer learning. To extract deep features, we employ pre-trained VGG and ResNet 50 architectures based on the ImageNet dataset, and output disease and pest images similar to a query image via a k-nearest-neighbor algorithm. In this study, we use a total of 23,868 images of 19 types of hot-pepper diseases and pests, for which, the proposed model achieves accuracies of 96.02 and 99.61%, respectively. We also measure the effects of fine-tuning and distance metrics. The results show that the use of fine-tuning-based deep features increases accuracy by approximately 0.7–7.38%, and the Bray–Curtis distance achieves an accuracy of approximately 0.65–1.51% higher than the Euclidean distance.

A Novel Fatigue Driving State Recognition and Warning Method Based on EEG and EOG Signals

Traffic accidents are easily caused by tired driving. If the fatigue state of the driver can be identified in time and a corresponding early warning can be provided, then the occurrence of traffic accidents could be avoided to a large extent. At present, the recognition of fatigue driving states is mostly based on recognition accuracy. Fatigue state is currently recognized by combining different features, such as facial expressions, electroencephalogram (EEG) signals, yawning, and the percentage of eyelid closure over the pupil over time (PERCLoS). The combination of these features increases the recognition time and lacks real-time performance. In addition, some features will increase error in the recognition result, such as yawning frequently with the onset of a cold or frequent blinking with dry eyes. On the premise of ensuring the recognition accuracy and improving the realistic feasibility and real-time recognition performance of fatigue driving states, a fast support vector machine (FSVM) algorithm based on EEGs and electrooculograms (EOGs) is proposed to recognize fatigue driving states. First, the collected EEG and EOG modal data are preprocessed. Second, multiple features are extracted from the preprocessed EEGs and EOGs. Finally, FSVM is used to classify and recognize the data features to obtain the recognition result of the fatigue state. Based on the recognition results, this paper designs a fatigue driving early warning system based on Internet of Things (IoT) technology. When the driver shows symptoms of fatigue, the system not only sends a warning signal to the driver but also informs other nearby vehicles using this system through IoT technology and manages the operation background.

Automatic modal parameters identification and uncertainty quantification based on block-bootstrap and multi-stage clustering under ambient excitation

This study proposes an algorithm for autonomous modal estimation to automatically eliminate false modes and quantify the uncertainty caused by the clustering algorithm and ambient factors. This algorithm belongs to the stochastic subspace identification (SSI) techniques and is based on the Block-Bootstrap and multi-stage clustering analysis. First, the Block-Bootstrap is introduced to decompose the response signal of the structure into M blocks of data. The covariance-driven stochastic subspace identification (SSI-Cov) method is used to process a random sample of data and obtain the corresponding M stabilization diagrams. In addition, the hierarchical clustering method is adopted to carry out the secondary clustering of the picked stable axis according to the defined distance threshold. Then, false modes are eliminated according to the proposed true and false modal discrimination index ( MDI). Finally, the above steps are repeated B times, and MDI is used to modify the initial modal parameters of group B. The mean value of elements in the cluster is taken as the recognition result of modal parameters, and the standard deviation is used to measure the accuracy of the recognition result. The numerical simulation results and the modal parameter identification of the Jing-yuan Yellow River Bridge show that, for identifying true and false modals, the proposed modal discrimination index is more effective than the threshold value of the traditional index. Also, it was found that the proposed method can eliminate the uncertainty introduced in the clustering process. In addition, this method can remove the influence of ambient noises, and it can improve the identification accuracy. It will be shown that this method has better anti-noise performance.

New Method for Adjusting Coefficients PI Conroller in the Load Control System of the Shearer

The paper proposes a method using extrapolator models to predict the parameters of the PI controller. With a sudden change in the resistance of coal to cutting, in particular, at different levels of steps and their combinations, the speed of the cutting current controller and current steps also differ, the quieting time of the transient process also occurs in different times. To implement the adjustment of the coefficients of the PI controller, a feed forward artificial neural network is used, which acts as an operational means of recognizing a multidimensional response curve in the control loop. Neural networks of two architectures were used: with a scalar and vector output function. The recognition result is the estimated level of the step in the coal’s resistance to cutting based on the initial counts of the multidimensional response curve at the specified coefficients of the PI controller. The approach proposed in the article can be used in solving other applied problems in which the control object is nonlinear and the control algorithm needs to adapt to the changing unobservable effect on the object.

An Online Cursive Handwritten Medical Words Recognition System for Busy Doctors in Developing Countries for Ensuring Efficient Healthcare Service Delivery

Doctors in developing countries are too busy to write digital prescriptions. Ninety-seven percent of Bangladeshi doctors write handwritten prescriptions, the majority of which lack legibility. Prescriptions are harder to read as they contain multiple languages. This paper proposes a machine learning approach to recognize doctors' handwriting to create digital prescriptions. A ‘Handwritten Medical Term Corpus’ dataset is developed containing 17,431 samples of 480 medical terms. In order to improve the recognition efficiency, this paper introduces a data augmentation technique to widen the variety and increase the sample size. A sequence of line data is extracted from the augmented images of 1,591,100 samples and fed to a Bidirectional LSTM. Data augmentation includes pattern Rotating, Shifting and Stretching (RSS). Eight different combinations are applied to evaluate the strength of the proposed method. The result shows 93.0% average accuracy (max: 94.5%, min: 92.1%) using Bidirectional LSTM and RSS data augmentation. This accuracy is 19.6% higher than the recognition result with no data expansion. The proposed handwritten recognition technology can be installed in a smartpen for busy doctors which will recognize the writings and digitize them in real-time. It is expected that the smartpen will contribute to reduce medical errors, save medical costs and ensure healthy living in developing countries.

Design of Multi-Receptive Field Fusion-Based Network for Surface Defect Inspection on Hot-Rolled Steel Strip Using Lightweight Dataset

With the advancement of industrial intelligence, defect recognition has become an indispensable part of facilitating surface quality in the steel manufacturing process. To assure product quality, most previous studies were typically trained with many defect samples. Nonetheless, a large quantity of defect samples is difficult to obtain, owing to the rare occurrence of defects. In general, deep learning-based methods underperformed as they have inherent limitations due to inadequate information, thereby restraining the application of models. In this study, a two-level Gaussian pyramid is applied to decompose raw data into different resolution levels simultaneously filtering the noises to acquire compact and representative features. Subsequently, a multi-receptive field fusion-based network (MRFFN) is developed to learn the hierarchical features and synthesize the respective prediction scores to form the final recognition result. As a result, the proposed method is capable of exhibiting an outstanding performance of 99.75% when trained using a lightweight dataset. In addition, the experiments conducted using the disturbance defect dataset showed the robustness of the proposed MRFFN against common noises and motion blur.

Gait-stride-and-frequency-based human intention recognition approach and experimental verification on lower limb exoskeleton

In the research of lower extremity exoskeleton, how to achieve synchronization between human and machine is quite significant. The intention recognition, which can be divided into three categories including EMG-based, EEG-based and biomechanics-based, is one of the effective implementation methods. In this paper, a new biomechanics-based method to realize the intention recognition is proposed. Compared with the mainstream, this method identifies the characteristic value of stride and frequency during walking, which describes human intention mathematically and concretizes the intention of human movement, improving the accuracy of recognition result and streamlining the algorithm. In addition, the impedance model is designed to further correct the recognition error. The main contents of this paper can be roughly summarized as follows. Gait feature event points are detected according to the angular signals of exoskeleton joints and the pressure signals of foot sole during the wearer’s walking process. Then the whole gait cycle is segmented by the identified gait feature event points, which is used to identify the wearer’s gait step and frequency in the gait cycle and output the trajectory transformed from standard gait trajectory by the recognized stride and frequency. Moreover, the interactive force signal collected by the three-dimensional force sensors mounted on the four-legged bar is provided as input to the designed impedance controller to adjust the transformed trajectory again. Also, the final trajectory is input to the Proportion Integral and Differential (PID) controller to realize the motion function of the lower extremity exoskeleton based on the wearer’s intention recognition result. Moreover, a simple hardware platform of lower limb exoskeleton is designed and built for practical experimental verification, which involves three kinds of gait respectively having constant stride, constant frequency and time-varying stride and frequency. The feasibility and reliability of the proposed algorithm can be concluded by analyzing the satisfactory experiment result.