|
|
|
题名
|
作者
|
年代
|
出处
|
被引量
|
| 1 | Crosstalk of liver immune cells and cell death mechanisms in different murine models of liver injury and its clinical relevance显示文摘BACKGROUND: Liver inflammation or hepatitis is a result of pluripotent interactions of cell death molecules, cytokines, chemokines and the resident immune cells collectively called as microenvironment. The interplay of these inflammatory mediators and switching of immune responses during hepatotoxic, viral, drug-induced and immune cell-mediated hepatitis decide the fate of liver pathology. The present review aimed to describe the mechanisms of liver injury, its relevance to human liver pathology and insights for the future therapeutic interventions.DATA SOURCES: The data of mouse hepatic models and relevant human liver diseases presented in this review are systematically collected from Pub Med, Science Direct and the Web of Science databases published in English. RESULTS: The hepatotoxic liver injury in mice induced by the metabolites of CCl4, acetaminophen or alcohol represent necrotic cell death with activation of cytochrome pathway, formation of reactive oxygen species(ROS) and mitochondrial damage. The Fas or TNF-α induced apoptotic liver injury was dependent on activation of caspases, release of cytochrome c and apoptosome formation. The Con A-hepatitis demonstrated the involvement of TRAIL-dependent necrotic/necroptotic cell death with activation of RIPK1/3. The α-Gal Cer-induced liver injury was mediated by TNF-α. The LPS-induced hepatitis involved TNF-α, Fas/Fas L, and perforin/granzyme cell death pathways. The MHV3 or Poly(I:C) induced liver injury was mediated by natural killer cells and TNF-α signaling. The necrotic ischemia-reperfusion liver injury was mediated byhypoxia, ROS, and pro-inflammatory cytokines; however, necroptotic cell death was found in partial hepatectomy. The crucial role of immune cells and cell death mediators in viral hepatitis(HBV, HCV), drug-induced liver injury, non-alcoholic fatty liver disease and alcoholic liver disease in human were discussed.CONCLUSIONS: The mouse animal models of hepatitis provide a parallel approach for the study of human liver pathology. Blocking or stimulating the pathways associated with liver cell death could unveil the novel therapeutic strategies in the management of liver diseases. | Hilal Ahmad Khan Muhammad Zishan Ahmad Junaid Ali Khan Muhammad Imran Arshad | 2017 | Hepatobiliary & Pancreatic Diseases International2017，16，3： | 24 |
| 2 | Numerical evaluation of new Austrian tunneling method excavation sequences: A case study显示文摘The main aspects that require attention in tunnel design in terms of safety and economy are the precise estimation of probable ground conditions and ground behavior during construction. The variation in rock mass behavior due to tunnel excavation sequence plays an important role during the construction stage.The purpose of this research is to numerically evaluate the effect of excavation sequence on the ground behavior for the Lowari tunnel project, Pakistan. For the tunnel stability, the ground behavior observed during the actual partial face excavation sequence is compared with the top heading and bench excavation sequence. For this purpose, the intact rock parameters are used along with the characterization of rock mass joints related parameters to provide input for numerical modelling via FLAC 2D. The in-situ stresses for the numerical modelling are obtained using empirical equations. From the comparison of the two excavation sequences, it was observed that the actual excavation sequence used for Lowari tunnel construction utilized more support than the top heading and bench method. However, the actual excavation sequence provided good results in terms of stability. | Hafeezur Rehman Abdul Muntaqim Naji Wahid Ali Muhammad Junaid Rini Asnida Abdullah Han-kyu Yoo | 2020 | International Journal of Mining Science and Technology2020，30，3： | 2 |
| 3 | COVID-19 and comorbidities of hepatic diseases in a global perspective显示文摘The worldwide outbreak of coronavirus disease 2019(COVID-19) has challenged the priorities of healthcare system in terms of different clinical management and infection transmission, particularly those related to hepatic-disease comorbidities. Epidemiological data evidenced that COVID-19 patients with altered liver function because of hepatitis infection and cholestasis have an adverse prognosis and experience worse health outcomes. COVID-19-associated liver injury is correlated with various liver diseases following a severe acute respiratory syndrome-coronavirus type 2(SARS-CoV-2) infection that can progress during the treatment of COVID-19 patients with or without pre-existing liver disease. SARS-CoV-2 can induce liver injury in a number of ways including direct cytopathic effect of the virus on cholangiocytes/hepatocytes, immune-mediated damage, hypoxia, and sepsis. Indeed, immediate cytopathogenic effects of SARSCoV-2 via its potential target, the angiotensin-converting enzyme-2 receptor, which is highly expressed in hepatocytes and cholangiocytes, renders the liver as an extra-respiratory organ with increased susceptibility to pathological outcomes. But, underlying COVID-19-linked liver disease pathogenesis with abnormal liver function tests(LFTs) is incompletely understood. Hence, we collated COVID-19-associated liver injuries with increased LFTs at the nexus of pre-existing liver diseases and COVID-19, and defining a plausible pathophysiological triad of COVID-19, hepatocellular damage, and liver disease. This review summarizes recent findings of the exacerbating role of COVID-19 in pre-existing liver disease and vice versa as well as international guidelines of clinical care, management, and treatment recommendations for COVID-19 patients with liver disease. | Aqsa Ahmad Syeda Momna Ishtiaq Junaid Ali Khan Rizwan Aslam Sultan Ali Muhammad Imran Arshad | 2021 | World Journal of Gastroenterology2021，27，13： | 1 |
| 4 | Cloud-based vs.blockchain-based IoT:a comparative survey and way forward显示文摘The Internet of Things(IoT)has been evolving for more than a decade.Technological advancements have increased its popularity,but concerns and risks related to IoT are growing considerably along with the increased number of connected devices.In 2013,a new cryptography-based infrastructure called blockchain emerged with the potential to replace the existing cloud-based infrastructure of IoT through decentralization.In this article,we provide a taxonomy of the challenges in the current IoT infrastructure,and a literature survey with a taxonomy of the issues to expect in the future of the IoT after adopting blockchain as an infrastructure.The two architectures are compared based on their strengths and weaknesses.Then a brief survey of ongoing key research activities in blockchain is presented,which will have considerable impact on overcoming the challenges encountered in the applicability of blockchain in IoT.Finally,considering the challenges and issues in both infrastructures and the latest research activities,we propose a high-level hybrid IoT approach that uses the cloud,edge/fog,and blockchain together to avoid the limitations of each infrastructure. | Raheel Ahmed MEMON Jian Ping LI Junaid AHMED Muhammad Irshad NAZEER Muhammad ISMAIL Khursheed ALI | 2020 | Frontiers of Information Technology & Electronic Engineering2020，21，4： | 1 |
| 5 | Novel Approach for a van der Pol Oscillator in the Continuous Time Domain显示文摘We investigate the continuous time domain numerical treatment of a van der Pol oscillator,applying the trial solution as an artificial feed-forward neural network model containing unknown adjustable parameters.The optimization of the network is performed by simulated annealing in an unsupervised method.The proposed scheme is tested successfully by its application in both non-stiff and stiff conditions.Its reliability and effectiveness is validated through comprehensive statistical analyses.The obtained results are in good agreement with the classical RK45 method. | Junaid Ali Khan Muhammad Asif Zahoor Raja Ijaz Mansoor Qureshi | 2011 | Chinese Physics Letters2011，28，11： | 1 |
| 6 | Stochastic Computational Approach for Complex Nonlinear Ordinary Differential Equations显示文摘We present an evolutionary computational approach for the solution of nonlinear ordinary differential equations(NLODEs).The mathematical modeling is performed by a feed-forward artificial neural network that defines an unsupervised error.The training of these networks is achieved by a hybrid intelligent algorithm,a combination of global search with genetic algorithm and local search by pattern search technique.The applicability of this approach ranges from single order NLODEs,to systems of coupled differential equations.We illustrate the method by solving a variety of model problems and present comparisons with solutions obtained by exact methods and classical numerical methods.The solution is provided on a continuous finite time interval unlike the other numerical techniques with comparable accuracy.With the advent of neuroprocessors and digital signal processors the method becomes particularly interesting due to the expected essential gains in the execution speed. | Junaid Ali Khan Muhammad Asif Zahoor Raja Ijaz Mansoor Qureshi | 2011 | Chinese Physics Letters2011，28，2： | 1 |
| 7 | Cardiopulmonary resuscitation: outcome and its predictors among hospitalized adult patients in Pakistan显示文摘 | Nadeem Ullah Khan Junaid A. Razzak Humaid Ahmed Muhammad Furqan Ali Faisal Saleem Hammad Alam Anwar ul Huda Uzma Rahim Khan Rifat Rehmani | 2008 | International Journal of Emergency Medicine2008，，1： | 1 |
| 8 | The Development Trends and Research Fronts in Orbital Angular Momentum Technology:A Bibliometric Analysis显示文摘Orbital angular momentum(OAM)technology,refers to Laguerre-Gaussian(LG)beams,twisted beams,vector/vortex beams,acoustic vortex beams and fractional vortex beams.It is an emerging and promising technology to improve the communication capacity,spectral efficiency,and anti-jamming capability due to its helical phase fronts and infinite orthogonal states.Although the OAM research began in the 1990s,the developing trends,current status,issues and characteristics through a systematic observation have not yet been performed.This paper presents a knowledge-based evolution of OAM research published in the Web of Science(WoS)from 2011 to 2021 using bibliometric analysis in Citepspace.The results demonstrate that the bandwidth,efficiency,gain,divergence,phase quantization,bulky and complex feeding structures,misalignment,distortion,interferences atmospheric turbulence and diffraction were the key issues found in the OAM technology.The main research hotspots and categories,influential authors,leading journals,best institutions of OAM show a strong bias in favor of their functions and technology developments.The research on OAM was mainly performed by the counties that have developed the 5G and now moving towards 6G communications like China,USA and South Korea.This study would serve as an inclusive guide on the future research trends and status especially for the OAM researchers. | Saleemullah Memon Xiuping Li Kamran Ali Memon Yuhan Huang Junaid Ahmed Uqaili Muhammad Ishfaq | 2023 | China Communications2023，20，2： | 0 |
| 9 | Citrus Diseases Recognition Using Deep Improved Genetic Algorithm显示文摘Agriculture is the backbone of each country,and almost 50%of the population is directly involved in farming.In Pakistan,several kinds of fruits are produced and exported the other countries.Citrus is an important fruit,and its production in Pakistan is higher than the other fruits.However,the diseases of citrus fruits such as canker,citrus scab,blight,and a few more impact the quality and quantity of this Fruit.The manual diagnosis of these diseases required an expert person who is always a time-consuming and costly procedure.In the agriculture sector,deep learning showing significant success in the last five years.This research work proposes an automated framework using deep learning and best feature selection for citrus diseases classification.In the proposed framework,the augmentation technique is applied initially by creating more training data from existing samples.They were then modifying the two pre-trained models named Resnet18 and Inception V3.The modified models are trained using an augmented dataset through transfer learning.Features are extracted for each model,which is further selected using Improved Genetic Algorithm(ImGA).The selected features of both models are fused using an array-based approach that is finally classified using supervised learning classifiers such as Support Vector Machine(SVM)and name a few more.The experimental process is conducted on three different datasets-Citrus Hybrid,Citrus Leaf,and Citrus Fruits.On these datasets,the best-achieved accuracy is 99.5%,94%,and 97.7%,respectively.The proposed framework is evaluated on each step and compared with some recent techniques,showing that the proposed method shows improved performance. | Usra Yasmeen Muhammad Attique Khan Usman Tariq Junaid Ali Khan Muhammad Asfand EYar ChAvais Hanif Senghour Mey Yunyoung Nam | 2022 | Computers, Materials & Continua2022，，5： | 0 |
| 10 | Deep-Net:Fine-Tuned Deep Neural Network Multi-Features Fusion for Brain Tumor Recognition显示文摘Manual diagnosis of brain tumors usingmagnetic resonance images(MRI)is a hectic process and time-consuming.Also,it always requires an expert person for the diagnosis.Therefore,many computer-controlled methods for diagnosing and classifying brain tumors have been introduced in the literature.This paper proposes a novel multimodal brain tumor classification framework based on two-way deep learning feature extraction and a hybrid feature optimization algorithm.NasNet-Mobile,a pre-trained deep learning model,has been fine-tuned and twoway trained on original and enhancedMRI images.The haze-convolutional neural network(haze-CNN)approach is developed and employed on the original images for contrast enhancement.Next,transfer learning(TL)is utilized for training two-way fine-tuned models and extracting feature vectors from the global average pooling layer.Then,using a multiset canonical correlation analysis(CCA)method,features of both deep learning models are fused into a single feature matrix—this technique aims to enhance the information in terms of features for better classification.Although the information was increased,computational time also jumped.This issue is resolved using a hybrid feature optimization algorithm that chooses the best classification features.The experiments were done on two publicly available datasets—BraTs2018 and BraTs2019—and yielded accuracy rates of 94.8%and 95.7%,respectively.The proposedmethod is comparedwith several recent studies andoutperformed inaccuracy.In addition,we analyze the performance of each middle step of the proposed approach and find the selection technique strengthens the proposed framework. | Muhammad Attique Khan Reham R.Mostafa Yu-Dong Zhang Jamel Baili Majed Alhaisoni Usman Tariq Junaid Ali Khan Ye Jin Kim Jaehyuk Cha | 2023 | Computers, Materials & Continua2023，76，9： | 0 |
| 11 | An Innovative Approach Utilizing Binary-View Transformer for Speech Recognition Task显示文摘The deep learning advancements have greatly improved the performance of speech recognition systems,and most recent systems are based on the Recurrent Neural Network(RNN).Overall,the RNN works fine with the small sequence data,but suffers from the gradient vanishing problem in case of large sequence.The transformer networks have neutralized this issue and have shown state-of-the-art results on sequential or speech-related data.Generally,in speech recognition,the input audio is converted into an image using Mel-spectrogram to illustrate frequencies and intensities.The image is classified by the machine learning mechanism to generate a classification transcript.However,the audio frequency in the image has low resolution and causing inaccurate predictions.This paper presents a novel end-to-end binary view transformer-based architecture for speech recognition to cope with the frequency resolution problem.Firstly,the input audio signal is transformed into a 2D image using Mel-spectrogram.Secondly,the modified universal transformers utilize the multi-head attention to derive contextual information and derive different speech-related features.Moreover,a feedforward neural network is also deployed for classification.The proposed system has generated robust results on Google’s speech command dataset with an accuracy of 95.16%and with minimal loss.The binary-view transformer eradicates the eventuality of the over-fitting problem by deploying a multiview mechanism to diversify the input data,and multi-head attention captures multiple contexts from the data’s feature map. | Muhammad Babar Kamal Arfat Ahmad Khan Faizan Ahmed Khan Malik Muhammad Ali Shahid Chitapong Wechtaisong Muhammad Daud Kamal Muhammad Junaid Ali Peerapong Uthansakul | 2022 | Computers, Materials & Continua2022，，9： | 0 |
| 12 | Modelling of debris-flow susceptibility and propagation: a case study from Northwest Himalaya显示文摘The geological and geographical position of the Northwest Himalayas makes it a vulnerable area for mass movements particularly landslides and debris flows. Mass movements have had a substantial impact on the study area which is extending along Karakorum Highway(KKH) from Besham to Chilas. Intense seismicity, deep gorges, steep terrain and extreme climatic events trigger multiple mountain hazards along the KKH, among which debris flow is recognized as the most destructive geohazard. This study aims to prepare a field-based debris flow inventory map at a regional scale along a 200 km stretch from Besham to Chilas. A total of 117 debris flows were identified in the field, and subsequently, a point-based debris-flow inventory and catchment delineation were performed through Arc GIS analysis. Regional scale debris flow susceptibility and propagation maps were prepared using Weighted Overlay Method(WOM) and Flow-R technique sequentially. Predisposing factors include slope, slope aspect, elevation, Topographic Roughness Index(TRI), Topographic Wetness Index(TWI), stream buffer, distance to faults, lithology rainfall, curvature, and collapsed material layer. The dataset was randomly divided into training data(75%) and validation data(25%). Results were validated through the Receiver Operator Characteristics(ROC) curve. Results show that Area Under the Curve(AUC) using WOM model is 79.2%. Flow-R propagation of debris flow shows that the 13.15%, 22.94%, and 63.91% areas are very high, high, and low susceptible to debris flow respectively. The propagation predicated by Flow-R validates the naturally occurring debris flow propagation as observed in the field surveys. The output of this research will provide valuable input to the decision makers for the site selection, designing of the prevention system, and for the protection of current infrastructure. | Hamza DAUD Javed Iqbal TANOLI Sardar Muhammad ASIF Muhammad QASIM Muhammad ALI Junaid KHAN Zahid Imran BHATTI Ishtiaq Ahmad Khan JADOON | 2024 | Journal of Mountain Science2024，21，1： | 0 |
| 13 | Crop Yield Prediction Using Machine Learning Approaches on a Wide Spectrum显示文摘The exponential growth of population in developing countries likeIndia should focus on innovative technologies in the Agricultural processto meet the future crisis. One of the vital tasks is the crop yield predictionat its early stage;because it forms one of the most challenging tasks inprecision agriculture as it demands a deep understanding of the growth patternwith the highly nonlinear parameters. Environmental parameters like rainfall,temperature, humidity, and management practices like fertilizers, pesticides,irrigation are very dynamic in approach and vary from field to field. In theproposed work, the data were collected from paddy fields of 28 districts in widespectrum of Tamilnadu over a period of 18 years. The Statistical model MultiLinear Regression was used as a benchmark for crop yield prediction, whichyielded an accuracy of 82% owing to its wide ranging input data. Therefore,machine learning models are developed to obtain improved accuracy, namelyBack Propagation Neural Network (BPNN), Support Vector Machine, andGeneral Regression Neural Networks with the given data set. Results showthat GRNN has greater accuracy of 97% (R2 = 0.97) with a normalizedmean square error (NMSE) of 0.03. Hence GRNN can be used for crop yieldprediction in diversified geographical fields. | SVinson Joshua ASelwin Mich Priyadharson Raju Kannadasan Arfat Ahmad Khan Worawat Lawanont Faizan Ahmed Khan Ateeq Ur Rehman Muhammad Junaid Ali | 2022 | Computers, Materials & Continua2022，，9： | 0 |
| 14 | Brain Tumor Segmentation using Multi-View Attention based Ensemble Network显示文摘Astrocytoma IV or glioblastoma is one of the fatal and dangerous types of brain tumors.Early detection of brain tumor increases the survival rate and helps in reducing the fatality rate.Various imaging modalities have been used for diagnosing by expert radiologists,and Medical Resonance Image(MRI)is considered a better option for detecting brain tumors as MRI is a non-invasive technique and provides better visualization of the brain region.One of the challenging issues is to identify the tumorous region from the MRI scans correctly.Manual segmentation is performed by medical experts,which is a time-consuming task and got chances of errors.To overcome this issue,automatic segmentation is performed for quick and accurate results.The proposed approach is to capture inter-slice information and reduce the outliers.Deep learning-based brain tumor segmentation techniques proved best among available segmentation techniques.However,deep learning may miss some preliminary info while using MRI images during segmentation.As MRI volumes are volumetric,3D U-Net-based models are used but complex.Combinations of multiple 2D U-Net predictions in axial,sagittal,and coronal views help to capture inter-slice information.This approach may reduce the system complexity.Moreover,the Conditional Random Fields(CRF)reduce the predictions’false positives and improve the segmentation results.This model is applied to Brain Tumor Segmentation(BraTS)2019 dataset,and cross-validation is performed to check the accuracy of results.The proposed approach achieves Dice Similarity Score(DSC)of 0.77 on Enhancing Tumor(ET),0.90 on Whole Tumor(WT),and 0.84 on Tumor Core(TC)with reduced Hausdorff Distance(HD)of 3.05 on ET,5.12 on WT and 3.89 on TC. | Noreen Mushtaq Arfat Ahmad Khan Faizan Ahmed Khan Muhammad Junaid Ali Malik Muhammad Ali Shahid Chitapong Wechtaisong Peerapong Uthansakul | 2022 | Computers, Materials & Continua2022，，9： | 0 |
| 15 | Graphene Nanosheets Decorated with Copper Oxide Nanoparticles for the Photodegradation of Methylene Blue显示文摘Textile industries extensively use colorants,such as methylene blue,and if disposed off untreated,they contaminate the effluent streams,causing a severe impact on the environment and aquatic life.Photocatalytic degradation has been found as an inevitable approach to treat them.Herein,we decorated the copper oxide nanoparticles on graphene nanosheets during the reflux process.The resultant copper oxide/graphene nanocomposites were analyzed for structural and functional attributes.It was observed that on increasing the copper oxide contents,the z-average size of the resultant nanocomposites decreased.The X-ray diffraction analysis demonstrated the crystalline nature of the nanocomposite.The surface morphology of the copper oxide nanoparticles appeared to be spherical and that of the copper oxide/graphene composite somehow wrinkled.The infrared analysis indicated successful intercalation of precursors in the nanocomposite.The bandgap of copper oxide/graphene nanocomposites varied in the range of 1.03—1.30 eV,which indicated their effective photocatalytic activity.The results demonstrated that after 120 min of exposure,the methylene blue removal efficiency reached 94.0%,92.2%,and 89.4%(mass fraction)on the copper oxide/graphene nanocomposite at copper oxide nanoparticles to graphene nanosheets ratios of 1:1,1.5:1,and 2:1(mass ratio),respectively.The photodegradation performance of the prepared nano-catalyst was found satisfactory even after five cycles. | Samavia RAFIQ Zulfiqar Ali RAZA Muhammad ASLAM Muhammad Junaid BAKHTIYAR | 2022 | Chemical Research in Chinese Universities2022，38，6： | 0 |
| 16 | Knee Osteoarthritis Classification Using X-Ray Images Based on Optimal Deep Neural Network显示文摘X-Ray knee imaging is widely used to detect knee osteoarthritis due to ease of availability and lesser cost.However,the manual categorization of knee joint disorders is time-consuming,requires an expert person,and is costly.This article proposes a new approach to classifying knee osteoarthritis using deep learning and a whale optimization algorithm.Two pre-trained deep learning models(Efficientnet-b0 and Densenet201)have been employed for the training and feature extraction.Deep transfer learning with fixed hyperparameter values has been employed to train both selected models on the knee X-Ray images.In the next step,fusion is performed using a canonical correlation approach and obtained a feature vector that has more information than the original feature vector.After that,an improved whale optimization algorithm is developed for dimensionality reduction.The selected features are finally passed to the machine learning algorithms such as Fine-Tuned support vector machine(SVM)and neural networks for classification purposes.The experiments of the proposed framework have been conducted on the publicly available dataset and obtained the maximum accuracy of 90.1%.Also,the system is explained using Explainable Artificial Intelligence(XAI)technique called occlusion,and results are compared with recent research.Based on the results compared with recent techniques,it is shown that the proposed method’s accuracy significantly improved. | Abdul Haseeb Muhammad Attique Khan Faheem Shehzad Majed Alhaisoni Junaid Ali Khan Taerang Kim Jae-Hyuk Cha | 2023 | Computer Systems Science & Engineering2023，47，11： | 0 |
| 17 | Facile NiCo_(2)S_(4)/C nanocomposite: an efficient material for water oxidation显示文摘The water oxidation in alkaline media is a kinetically sluggish process and it requires an active electrocatalyst for overall water splitting which is a challenging task to date.Herein,we formulate a platform for the design of efficient NiCo_(2)S_(4)/C nanocomposite using earth abundant and nonprecious materials.The nanocomposites are prepared by scale up hydrothermal method using different carbon contents from acid dehydrated sucrose.They are structurally and morphologically character-ized by various analytic techniques.The scanning electron microscopy has shown few microns flower-like morphology of nanocomposite and hexagonal crystalline phase is identified by X-ray diffraction(XRD).Further,high-resolution transmission electron microscopy supported the XRD results,and C,Ni,Co and O elements were found in the composition nanocomposite as investigated by energy-dispersive spectroscopy.The most active nanocomposite reaches a current density of 20 mA·cm^(−2) at potential of 285 mV vs reversible hydrogen electrode.The nanocomposite is kinetically supported by 61 mV·dec^(−1) as small Tafel slope.The nanocomposite is stable and durable for 40 h.The electrochemical impedance spectroscopy described a small charge transfer resistance of 188.4Ω.These findings suggest that the NiCo_(2)S_(4)/C nanocomposite could be used as a promising material for an extended range of applications particularly in energy technology. | Umair Aftab Aneela Tahira Raffaello Mazzaro Vittorio Morandi Muhammad Ishaq Abro Muhammad Moazam Baloch Junaid Ali Syed Ayman Nafady Zafar Hussain Ibupoto | 2020 | Tungsten2020，2，4： | 0 |
| 18 | Large Scale Fish Images Classification and Localization using Transfer Learning and Localization Aware CNN Architecture显示文摘Building an automatic fish recognition and detection system for largescale fish classes is helpful for marine researchers and marine scientists because there are large numbers of fish species.However,it is quite difficult to build such systems owing to the lack of data imbalance problems and large number of classes.To solve these issues,we propose a transfer learning-based technique in which we use Efficient-Net,which is pre-trained on ImageNet dataset and fine-tuned on QuT Fish Database,which is a large scale dataset.Furthermore,prior to the activation layer,we use Global Average Pooling(GAP)instead of dense layer with the aim of averaging the results of predictions along with having more information compared to the dense layer.To check the validity of our model,we validate our model on the validation set which achieves satisfactory results.Also,for the localization task,we propose an architecture that consists of localization aware block,which captures localization information for better prediction and residual connections to handle the over-fitting problem.Actually,the residual connections help the layer to combine missing information with the relevant one.In addition,we use class weights and Focal Loss(FL)to handle class imbalance problems along with reducing false predictions.Actually,class weights assign less weights to classes having fewer instances and large weights to classes having more number of instances.During the localization,the qualitative assessment shows that we achieve 57%Mean Intersection Over Union(IoU)on testing data,and the classification results show 75%precision,70%recall,78%accuracy and 74%F1-Score for 468 fish species. | Usman Ahmad Muhammad Junaid Ali Faizan Ahmed Khan Arfat Ahmad Khan ArifUr Rehman Malik Muhammad Ali Shahid Mohd Anul Haq Ilyas Khan Zamil SAlzamil Ahmed Alhussen | 2023 | Computer Systems Science & Engineering2023，45，5： | 0 |
您所在的机构暂未开通此部分的功能，请联系管理员