Context-based segmentation of the longissimus muscle in beef with a deep neural network

The problem of segmenting the cross-section through the longissimus muscle in beef carcasses with computer vision methods was investigated. The available data were 111 images of cross-sections coming from 28 cows (typically four images per cow). Training data were the pixels of the muscles, marked manually. The AlexNet deep convolutional neural network was used as the classifier, and single pixels were the classified objects. Each pixel was presented to the network together with its small circular neighbourhood, and with its context represented by the further neighbourhood, darkened by halving the image intensity. The average classification accuracy was 96\%. The accuracy without darkening the context was found to be smaller, with a small but statistically significant difference. The segmentation of the longissimus muscle is the introductory stage for the next steps of assessing the quality of beef for the alimentary purposes.

Download Full-text

Real-Time Automated Classification of Sky Conditions Using Deep Learning and Edge Computing

Remote Sensing ◽

10.3390/rs13193859 ◽

2021 ◽

Vol 13 (19) ◽

pp. 3859

Author(s):

Joby M. Prince Czarnecki ◽

Sathishkumar Samiappan ◽

Meilun Zhou ◽

Cary Daniel McCraine ◽

Louis L. Wasson

Keyword(s):

Neural Network ◽

Deep Learning ◽

Image Quality ◽

Convolutional Neural Network ◽

Precision Agriculture ◽

Edge Computing ◽

Training Data ◽

Learning Approaches ◽

Sky Conditions

The radiometric quality of remotely sensed imagery is crucial for precision agriculture applications because estimations of plant health rely on the underlying quality. Sky conditions, and specifically shadowing from clouds, are critical determinants in the quality of images that can be obtained from low-altitude sensing platforms. In this work, we first compare common deep learning approaches to classify sky conditions with regard to cloud shadows in agricultural fields using a visible spectrum camera. We then develop an artificial-intelligence-based edge computing system to fully automate the classification process. Training data consisting of 100 oblique angle images of the sky were provided to a convolutional neural network and two deep residual neural networks (ResNet18 and ResNet34) to facilitate learning two classes, namely (1) good image quality expected, and (2) degraded image quality expected. The expectation of quality stemmed from the sky condition (i.e., density, coverage, and thickness of clouds) present at the time of the image capture. These networks were tested using a set of 13,000 images. Our results demonstrated that ResNet18 and ResNet34 classifiers produced better classification accuracy when compared to a convolutional neural network classifier. The best overall accuracy was obtained by ResNet34, which was 92% accurate, with a Kappa statistic of 0.77. These results demonstrate a low-cost solution to quality control for future autonomous farming systems that will operate without human intervention and supervision.

Download Full-text

Deep Learning Based Adaptive Recurrent Neural Network for Detection of Myocardial Infarction

Journal of Medical Imaging and Health Informatics ◽

10.1166/jmihi.2021.3913 ◽

2021 ◽

Vol 11 (12) ◽

pp. 3044-3053

Author(s):

Rakesh Kumar Mahendran ◽

V. Prabhu ◽

V. Parthasarathy ◽

A. Mary Judith

Keyword(s):

Neural Network ◽

Myocardial Infarction ◽

Deep Learning ◽

Recurrent Neural Network ◽

Classification Accuracy ◽

Learning Algorithm ◽

Training Data ◽

Discrete Wavelet ◽

Ecg Signal ◽

Detection Techniques

Myocardial infarction (MI) may precipitate severe health damage and lead to irreversible death of the heart muscle, the result of prolonged lack of oxygen if it is not treated in a timely manner. Lack of accurate and early detection techniques for this heart disease has reduced the efficiency of MI diagnosis. In this paper, the design, and implementation of an efficient deep learning algorithm called Adaptive Recurrent neural network (ARNN) is proposed for the MI detection. The main objective of the proposed work is the accurate identification of MI disease using ECG signals. ECG signal denoising has been performed using the Multi-Notch filter, which removes the specified noise frequency range. Discrete wavelet transform (DWT) is utilized for performing the feature extraction that decomposes the ECG signal into varied scales with waveletfiltering bank. After the extraction of specific QRS features, classification of the defected and normal ECG arrhythmic beat has been performed using the deep learning-based ARNN classifier. The MIT-BIH database has been used for testing and training data. The performance of the proposed algorithm is evaluated based on classification accuracy. Results that are attained include the classification accuracy of about 99.21%, 99% of sensitivity and 99.4% of specificity with PPV and NPV of about 99.4 and 99.01 values indicate the enhanced performance of our proposed work compared with the conventional LSTM-CAE and LSTM-CNN techniques.

Download Full-text

Analysis of COVID-19 Infections on a CT Image Using DeepSense Model

Frontiers in Public Health ◽

10.3389/fpubh.2020.599550 ◽

2020 ◽

Vol 8 ◽

Author(s):

Adil Khadidos ◽

Alaa O. Khadidos ◽

Srihari Kannan ◽

Yuvaraj Natarajan ◽

Sachi Nandan Mohanty ◽

...

Keyword(s):

Neural Network ◽

Deep Learning ◽

Input Data ◽

Training Data ◽

Learning Framework ◽

Learning Classifiers ◽

Tomography Image ◽

Mining Model ◽

Deep Learning Model

In this paper, a data mining model on a hybrid deep learning framework is designed to diagnose the medical conditions of patients infected with the coronavirus disease 2019 (COVID-19) virus. The hybrid deep learning model is designed as a combination of convolutional neural network (CNN) and recurrent neural network (RNN) and named as DeepSense method. It is designed as a series of layers to extract and classify the related features of COVID-19 infections from the lungs. The computerized tomography image is used as an input data, and hence, the classifier is designed to ease the process of classification on learning the multidimensional input data using the Expert Hidden layers. The validation of the model is conducted against the medical image datasets to predict the infections using deep learning classifiers. The results show that the DeepSense classifier offers accuracy in an improved manner than the conventional deep and machine learning classifiers. The proposed method is validated against three different datasets, where the training data are compared with 70%, 80%, and 90% training data. It specifically provides the quality of the diagnostic method adopted for the prediction of COVID-19 infections in a patient.

Download Full-text

Quality Index of Supervised Data for Convolutional Neural Network-Based Localization

Applied Sciences ◽

10.3390/app9101983 ◽

2019 ◽

Vol 9 (10) ◽

pp. 1983 ◽

Cited By ~ 2

Author(s):

Seigo Ito ◽

Mineki Soga ◽

Shigeyoshi Hiratsuka ◽

Hiroyuki Matsubara ◽

Masaru Ogawa

Keyword(s):

Neural Network ◽

Convolutional Neural Network ◽

Quality Index ◽

Single Photon ◽

Automated Guided Vehicles ◽

Training Data ◽

Supervised Training ◽

Localization Performance ◽

Imaging Sensor

Automated guided vehicles (AGVs) are important in modern factories. The main functions of an AGV are its own localization and object detection, for which both sensor and localization methods are crucial. For localization, we used a small imaging sensor named a single-photon avalanche diode (SPAD) light detection and ranging (LiDAR), which uses the time-of-flight principle and arrays of SPADs. The SPAD LiDAR works both indoors and outdoors and is suitable for AGV applications. We utilized a deep convolutional neural network (CNN) as a localization method. For accurate CNN-based localization, the quality of the supervised data is important. The localization results can be poor or good if the supervised training data are noisy or clean, respectively. To address this issue, we propose a quality index for supervised data based on correlations between consecutive frames visualizing the important pixels for CNN-based localization. First, the important pixels for CNN-based localization are determined, and the quality index of supervised data is defined based on differences in these pixels. We evaluated the quality index in indoor-environment localization using the SPAD LiDAR and compared the localization performance. Our results demonstrate that the index correlates well to the quality of supervised training data for CNN-based localization.

Download Full-text

Formation of ssb, dsb, and Uracil in Monofilarly and Bifilarly Bromouracil-Substituted DNA Molecules

Zeitschrift für Naturforschung C ◽

10.1515/znc-1992-3-415 ◽

1992 ◽

Vol 47 (3-4) ◽

pp. 255-263

Author(s):

Henning Menke ◽

Wolfgang Köhnlein

Keyword(s):

Cross Section ◽

Cross Sections ◽

Uv Light ◽

Linear Part ◽

Reversed Phase ◽

Direct Production ◽

Significant Difference ◽

Effect Curve ◽

The Cross

Abstract Bifilarly BU-substituted ColE 1 plasmid and monofilarly BU-substituted M 13 phage DNA were irradiated with UV light of 313 nm. Using agarose gel electrophoresis and “reversed phase” HPLC technique ssb, dsb induction and uracil formation, respectively, could be detected in the irradiated DNA in dependence on the UV fluence. The analysis of the strandbreaks in bifilar Col E1 DNA shows a significant part of directly induced dsb. Cross sections of ssb induction from 4.1 m2/J x 107 in 28%, 3.9 m2/J x 107 in 55% and 3.1 m2/J x 107 in 8 5 -9 0% BU-substituted DNA were calculated. The cross section for dsb induction was found to be 0.04 m2/J x 107, estimated from the linear part of the fluence effect curve. In monofilar M13 DNA a linear fluence effect curve for dsb induction was obtained. Excluding other than the direct production of dsb by using an in vitro approach for M13 DNA , the results strongly support the hypothesis that dsb can be induced by one photochem ical absorption event. The cross section for ssb was 3.8 m2/J x 107 and for dsb 0.05 m2/J x 107 in 41.5% monofilarly BU-substituted M13 DNA . The comparison of ssb, dsb, and uracil production in bifilar and monofilar DNA with similar BU substitution showed no significant difference between the two DNA systems (ColE 1, M 13), indicating that the location of BU molecules in one or in both DNA strands will not lead to a different number of lesions after UV313 exposure.

Download Full-text

Deep Neural Network Algorithm Feedback Model with Behavioral Intelligence and Forecast Accuracy

Symmetry ◽

10.3390/sym12091465 ◽

2020 ◽

Vol 12 (9) ◽

pp. 1465

Author(s):

Taikyeong Jeong

Keyword(s):

Neural Network ◽

Classification Accuracy ◽

Deep Neural Network ◽

Time Series Data ◽

Training Data ◽

Series Data ◽

Context Aware ◽

Data Set ◽

Feedback Model ◽

Data Connection

When attempting to apply a large-scale database that holds the behavioral intelligence training data of deep neural networks, the classification accuracy of the artificial intelligence algorithm needs to reflect the behavioral characteristics of the individual. When a change in behavior is recognized, that is, a feedback model based on a data connection model is applied, an analysis of time series data is performed by extracting feature vectors and interpolating data in a deep neural network to overcome the limitations of the existing statistical analysis. Using the results of the first feedback model as inputs to the deep neural network and, furthermore, as the input values of the second feedback model, and interpolating the behavioral intelligence data, that is, context awareness and lifelog data, including physical activities, involves applying the most appropriate conditions. The results of this study show that this method effectively improves the accuracy of the artificial intelligence results. In this paper, through an experiment, after extracting the feature vector of a deep neural network and restoring the missing value, the classification accuracy was verified to improve by about 20% on average. At the same time, by adding behavioral intelligence data to the time series data, a new data connection model, the Deep Neural Network Feedback Model, was proposed, and it was verified that the classification accuracy can be improved by about 8 to 9% on average. Based on the hypothesis, the F (X′) = X model was applied to thoroughly classify the training data set and test data set to present a symmetrical balance between the data connection model and the context-aware data. In addition, behavioral activity data were extrapolated in terms of context-aware and forecasting perspectives to prove the results of the experiment.

Download Full-text

Digital Image Based Identification of Rice Variety Using Image Processing and Neural Network

TELKOMNIKA Indonesian Journal of Electrical Engineering ◽

10.11591/tijee.v16i1.1602 ◽

2015 ◽

Vol 16 (1) ◽

pp. 182

Author(s):

Lilik Sumaryanti ◽

Aina Musdholifah ◽

Sri Hartati

Keyword(s):

Neural Network ◽

Image Processing ◽

Classification Accuracy ◽

Rice Variety ◽

Texture Features ◽

Rice Varieties ◽

Food Grains ◽

Average Accuracy ◽

Neural Network Algorithm

The increased of consumer concern on the originality of rice variety and the quality of rice leads to originality certification of rice by existing institutions. Technology helps human to perform evaluations of food grains using images of objects. This study developed a system used as a tool to identify rice varieties. Identification process was performed by analyzing rice images using image processing. The analyzed features for identification consisted of six color features, four morphological features, and two texture features. Classifier used LVQ neural network algorithm. Identification results using a combination of all features gave average accuracy of 70,3% with the highest classification accuracy level of 96,6% for Mentik Wangi and the lowest classification accuracy of 30% for Cilosari.

Download Full-text

Identification of Rice Variety Using Geometric Features and Neural Network

IJCCS (Indonesian Journal of Computing and Cybernetics Systems) ◽

10.22146/ijccs.48203 ◽

2019 ◽

Vol 13 (3) ◽

pp. 301

Author(s):

Wahyu Srimulyani ◽

Aina Musdholifah

Keyword(s):

Neural Network ◽

Rice Variety ◽

Physical Characteristics ◽

Training Data ◽

Test Results ◽

Trial And Error ◽

Rice Varieties ◽

Segmentation Evaluation ◽

The Neural Network

Indonesia has many food varieties, one of which is rice varieties. Each rice variety has physical characteristics that can be recognized through color, texture, and shape. Based on these physical characteristics, rice can be identified using the Neural Network. Research using 12 features has not optimal results. This study proposes the addition of geometry features with Learning Vector Quantization and Backpropagation algorithms that are used separately.The trial uses data from 9 rice varieties taken from several regions in Yogyakarta. The acquisition of rice was carried out using a camera Canon D700 with a kit lens and maximum magnification, 55 mm. Data sharing is carried out for training and testing, and the training data was sharing with the quality of the rice. Preprocessing of data was carried out before feature extraction with the trial and error thresholding process of segmentation. Evaluation is done by comparing the results of the addition of 6 geometry features and before adding geometry features.The test results show that the addition of 6 geometry features gives an increase in the value of accuracy. This is evidenced by the Backpropagation algorithm resulting in increased accuracy of 100% and 5.2% the result of the LVQ algorithm.

Download Full-text

Assessing the Impact of Expert Labelling of Training Data on the Quality of Automatic Classification of Lithological Groups Using Artificial Neural Networks

Applied Computer Systems ◽

10.2478/acss-2020-0016 ◽

2020 ◽

Vol 25 (2) ◽

pp. 145-152

Author(s):

Yan Kuchin ◽

Ravil Mukhamediev ◽

Kirill Yakunin ◽

Janis Grundspenkis ◽

Adilkhan Symagulov

Keyword(s):

Neural Network ◽

Automatic Classification ◽

Training Data ◽

Classification Problems ◽

Expert Opinions ◽

Input Parameters ◽

Artificial Neural ◽

Artificial Neural Network Ann ◽

The Impact

AbstractMachine learning (ML) methods are nowadays widely used to automate geophysical study. Some of ML algorithms are used to solve lithological classification problems during uranium mining process. One of the key aspects of using classical ML methods is causing data features and estimating their influence on the classification. This paper presents a quantitative assessment of the impact of expert opinions on the classification process. In other words, we have prepared the data, identified the experts and performed a series of experiments with and without taking into account the fact that the expert identifier is supplied to the input of the automatic classifier during training and testing. Feedforward artificial neural network (ANN) has been used as a classifier. The results of the experiments show that the “knowledge” of the ANN of which expert interpreted the data improves the quality of the automatic classification in terms of accuracy (by 5 %) and recall (by 20 %). However, due to the fact that the input parameters of the model may depend on each other, the SHapley Additive exPlanations (SHAP) method has been used to further assess the impact of expert identifier. SHAP has allowed assessing the degree of parameter influence. It has revealed that the expert ID is at least two times more influential than any of the other input parameters of the neural network. This circumstance imposes significant restrictions on the application of ANNs to solve the task of lithological classification at the uranium deposits.

Download Full-text

Comparison of different algorithms used for creating river terrain model based on the cross-sections.

10.5194/egusphere-egu2020-9790 ◽

2020 ◽

Author(s):

Luděk Bureš ◽

Radek Roub ◽

Petra Sychová

Keyword(s):

Cross Section ◽

Cross Sections ◽

Hydrodynamic Model ◽

Linear Interpolation ◽

Cross Sectional ◽

Terrain Model ◽

Model Based ◽

Interpolation Techniques ◽

The Cross

Various techniques can be used to create a river terrain model. The most common technique uses 3D bathymetric points distributed across the main channel. The terrain model is then created using common interpolation techniques. The quality of this terrain depends on the number of the measured points and their location.An alternative method may be an application of a set of cross-sections. Special interpolation algorithms are used for this purpose. These algorithms create new bathymetric points between two adjacent cross-sections that are located in a composite bathymetric network (CBN). Common interpolation techniques can be used to create a river terrain model. The advantage of this approach is a necessity of smaller dataset.We present a comparison of four different algorithms for creating a river terrain model based on measured cross-sections. The first algorithm (A1) adopts a method of linear interpolation to create CBN [1]. The second algorithm (A2) reshapes the cross-sections and then applies linear interpolation. This reshaping allows better take into the account the thalweg line [2]. The third algorithm (A3) uses cross-sectional reshaping and uses cubic hermit splines to create CBN [3]. The last algorithm (A4) &#160;implies the channel boundary and the thalweg line as additional inputs. Additional inputs define the shape of the newly created river channel [4].Three different distances among individual cross-sections were used for the performance tests (50, 100 and 200 meters). The quality of topographic schematization and its impact on hydrodynamic model results were evaluated. Preliminary results show that there is almost no difference in the performance of the algorithms at cross-section distance of 50 m. The A4 algorithm outperforms/surpass its competitors in the case that input data (the cross-section distance is) are in 200 m spacing.This research was supported by the Operational Programme Prague &#8211; Growth Pole of the Czech Republic, project No. CZ.07.1.02/0.0/0.0/17_049/0000842, Tools for effective and safe management of rainwater in Prague city &#8211; RainPRAGUE.[1]&#160; &#160;&#160;&#160;&#160; Vetter, M., H&#246;fle, B., Mandelburger, G., Rutzinger, M. Estimating changes of riverine landscapes and riverbeds by using airborne LiDAR data and river cross-sections. Zeitschrift f&#252;r Geomorphologie, Supplementary Issues, 2011, 55.2: 51-65.[2]&#160;&#160;&#160;&#160;&#160;&#160; Chen, W., Liu, W. Modeling the influence of river cross-section data on a river stage using a two-dimensional /three-dimensional hydrodynamic model. Water, 2017, 9.3: 203.[3]&#160;&#160;&#160;&#160;&#160;&#160; Caviedes-Voulli&#232;me, D.; Morales-Hern&#225;ndez, M.; L&#243;pez-Marijuan, I.; Garc&#237;a-Navarro, P. Reconstruction of 2D river beds by appropriate interpolation of 1D cross-sectional information for flood simulation. Environ. Model. Softw., 2014, 61, 206&#8211;228.[4] &#160;&#160;&#160;&#160;&#160; Merwade, V.; Cook, A.; Coonrod, J. GIS techniques for creating river terrain models for hydrodynamic modeling and flood inundation mapping. Environ. Model. Softw., 2008, 23, 1300&#8211;1311.

Download Full-text