Machine Learning Framework for Face Masks Detection in Covid-19 Pandemic – A Survey
1Jose Triny K, 2Kavinkumar A, 3Nivash M, 4Pavithiran UK, 5Prabhu M
1,2,3,4,5
Department of computer science and Engineering, M.Kumarasamy College of Engineering, Thalavapalayam-Karur-639113.
1[email protected], 2[email protected], 3[email protected],
4[email protected], 5[email protected]
Abstract
The design of carrying confront veils freely is developing much appreciated to the COVID- 19 corona virus scourge all circular the world. Sometime recently Covid-19, people want to set on veils to ensure their wellness from discuss toxins. though diverse family unit self- cognizant with respect to their appearance, they cowl their feelings from the ultimate open with the assistance of concealment their faces. Researchers treated that wearing confront veils works on preventive COVID-19 transmission. COVID-19 (called corona virus) is that the today's scourge infection that hit the human wellbeing inside the closing century. In 2020, the expedient spreading of COVID-19 has compelled the globe Wellbeing Organization to announce COVID-19 as a worldwide widespread. Steady with over 5 million cases are aroused through COVID-19 in but half dozen months over 188 worldwide areas. The infection spreads by means of closed bit and in stuffed and packed districts. The corona virus COVID-19 widespread is delivering a world wellbeing fiasco. One among the successful security strategies is donning a confront veils freely zones in step with the globe Wellbeing Organization (WHO). Amid this overview, look at changed confront location and modernity strategies to spot the cover in unpracticed approach.
Index Terms:
Corona virus (COVID), World health organization, Machine learning method, Pandemic.
Introduction
Corona virus sickness 2019 (COVID-19) is Relate in Nursing irresistible clutter that started to multiply from city China, in Gregorian calendar month 2019. Interior a fast period, this sickness is desolated each corner of the field and so the World Wellbeing Organization pronounced this sickness as an flare-up on eleven Walk 2020. This sickness is much appreciated to the weight of Extreme Intense digestion system Disorder Corona virus two (SARS-CoV-2). In Gregorian calendar month 2020, cases come to about twelve million world, and misfortune of life much obliged to this disorder unbroken developing a day, and thus the passingontollis562,039.
Attentive the information and homes of COVID-19 it could be pronounced that this existence- threatening infection will unroll from character to individual through hack, sniffling, or indeed near to contact. As a result, it's returns to be exceptionally vital to return over the influenced people prior and disconnect them to anticipate any spreading of this infection.
Within the setting of transmitted virus between people with the assistance of sputtering (splashing), wearing the veil on the confront appears fundamental to ensure people and to
confinement the proliferation of the clutter. As of now, we're overseeing the 2019–2020 corona virus wide spread. Corona virus sickness 2019 (COVID-19) is relate irresistible disturbed with 1st signs exceptionally comparative to the grippe. The indication of COVID-19 was articulated 1st in China and in no time spreads to the consolation of the platform. The COVID- 19 contagiousness is known to be tall by means of comparison with the grippe. Amid this paper, we have a inclination to propose a fashion of a cell software bundle permitting individuals.
In spite of the fact that numerous people are as of now persuaded of the side interest for wearing confront assurance covers that grasp suggested by means of the planet Wellbeing Organization and clinical ponders, you'll be prepared to have a look at that a parcel of individuals do not with productivity set on their veils. These perceptions have intersection rectifier medical caretakers and elective voters to start bar campaigns related to the common open wellbeing tutoring in wearing the cover. Accurately, these campaigns contain sensitizing masses around the culminate and erroneous much obliged to wear confront security veils by diffusive bar blurbs and drawings. Amid this state of issues, current machine exhorts to direct those open wellbeing campaigns by strategy of arranging relate degree photo essentially based relate degree investigation procedure and an associated virtual gadget that are committed to the confirmation of the correct veil carrying by misusing a remote phone and its frontal camera. The essential layout is shown in fig1.
Fig 1: Face Mask Detection.
Related Work
[1] It given a near examination of all the central parts related with the COVID-19 widespread.
Furthermore, to the coordinate wellness suggestions related with the irruption of COVID-19, this has a look at highlights its effect on the around the world national economy. In drawing things to a near to, we find the business of advances which has the {web the net} of Things (IoT), unmanned Airborne Vehicles (UAVs), blockchain, computer science (AI), and 5G, among others, to help moderate the effect of COVID-19 irruption.
[2, 11] It attempted to benefit logical specialists and clinical analysts in overpowering their gone up against issues whereas overseeing COVID-19 tremendous measurements. The explored strategies set forward progresses in logical measurements investigation with relate exactitude of the greatest sum as 98. we have a propensity to furthermore come back to be with relate full talk with respect to be that as it may AI execution may be a curiously large advantage in combating different comparable infections. It's what is more accommodating to empower the investigation of this contamination by making utilize of analyzing the open
measurements. Reenacted insights will help in making suitable treatment regimens, neutralization strategies, and medicament and inoculation headway.
[3, 32, 45] It upheld a cross breed demonstrate exploitation profound and classical comfort finding out for veil location is additionally advertised. The arranged show incorporates parts.
The essential issue is planning for work extraction the business of Resnet50. Though the other issue is planning for the classification strategy of confront covers exploitation choice trees, Bolster Vector Machine (SVM), and gathering set of rules. 3 confront masked datasets were chosen for investigation. The 3 datasets unit the Real-World disguised Face Dataset (RMFD).
The Recreated masked Confront Dataset (SMFD), conjointly the labeled Faces inside the Wild (LFW).
[4, 29, 48, 50] It examined a few Profound Learning (DL) strategies had been outlined to succeed in this point, which has Generative Ill-disposed Systems (GANs), Extraordinary Learning Machine (ELM), and Long /Brief Term Memory (LSTM). It portrays relate encased bioinformatics strategy inside which one in all a shape perspective of insights from a time of built up and unstructured actualities assets unit put along to make the simple structures for doctors and analysts. The fundamental great thing around these AI-based through and through frameworks is to zest up the strategy of ponder and treatment of the COVID-19 ailment. The first later related distributions and logical reports had been examined with the clarification of selecting inputs and targets of the community that will encourage coming to a solid Counterfeit Neural Network-primarily based mostly device for challenges related to COVID-19. What is more, there square measure a few unequivocal inputs for each stage, comprising of different.
[5, 35, 39, 40, 41] It arranged relate ACGAN based for the most part demonstrate expressed as CovidGAN that creates manufactured CXR pics to expand the dataset and to boost the execution of CNN in COVID-19 location. The consider unit of measurement connected on a dataset with 403COVID-CXR snap shots and 721 Normal-CXR photographs. Our kept dataset highlights the deficiency of therapeutic pics inside the investigation groups. At first, the arranged CNN plan is employed to classify the two categories (this can be COVID-CXR and Normal-CXR). Advance, the execution of CNN with counterfeit records increase procedure is explored. Fake information augmentation provides extra inconstancy to the dataset, through extending it. CovidGAN is utilized to induce manufactured photographs of chest X-ray (CXR).
[6, 27, 28, 49] It arranged relate ACGAN based for the most part demonstrate expressed as CovidGAN that creates manufactured CXR pics to expand the dataset and to boost the execution of CNN in COVID-19 location. The consider unit of measurement connected on a dataset with 403COVID-CXR snap shots and 721 Normal-CXR photographs. Our kept dataset highlights the deficiency of therapeutic pics inside the investigation groups. At first, the arranged CNN plan is employed to classify the two categories (this can be COVID-CXR and Normal-CXR). Advance, the execution of CNN with counterfeit records increase procedure is explored. Fake information augmentation provides extra inconstancy to the dataset, through extending it. CovidGAN is utilized to induce manufactured photographs of chest X-ray (CXR).
[7, 33] It arranged a haggard directed profound finding out methodology for police examination and classifying COVID-19 contamination from CT photographs. The arranged technique will minimize the needs of direct naming of CT pix but indeed so be able to harvest redress
defilement discovery and recognize COVID-19 from non-COVID-19 cases. Based generally at the promising results noninheritable subjectively and quantitatively, will conceive of an in profundity preparing of progressed approach in massive-scale clinical investigation.
[8, 24] It arranged a solitary on line consideration module with a three-D convolutional community (CNN) to target the defilement zones in lungs once making choices of analyze. Note that there exists unequal conveyance of the sizes of the defilement locales among COVID-19 and CAP, mostly due to fast advance of COVID-19 when indication onset. Hence, we have a propensity to extend a dual-sampling approach to relieve the lopsided acing. Our approach is assessed (to our most prominent information) upon the foremost vital multi-center CT.
[9, 12, 13] It arranged an awfully unused and particular strategy to adaptation the worldly energetic of facial expressions as profound directions on the SPD complex. As relate expansion of the course pipeline of fluctuation descriptors, we have a inclination to take after SVM with true blue huge particular bits determined from universal arrangement for profound change directions course.
[10, 14, 15, 16, 19, 20, 21, 23, 24] It laid out stays as photographs that entirely acclimate the
„„gravity run the show computerized camera‟‟, that takes after the presumption that camera facilitate of the vertical pivot have to be interminably be relentless therewith of the neighborhood head arrange. We have a inclination to define symbol combine as a result of the regularized picture made by implies that of TSIR aboard its grapple partner, each of which may well be nourished into the AGPE module for evaluating charming- grained head postures. This paper in expansion proposes relate Anchor-Guided Pair wise Misfortune (AGPL), that portrays the subordinate connectedness of postures between each combine of pictures. The arranged method is assessed and substantial with sufficient tests that appear its viability. Comprehensive tests appear that our approach outflanks the dynamic image-primarily based for the most part ways on each in door and out of entryways datasets.
Machine learning methods
Machine finding out (ML) is that the watch of pc calculations that upgrade consequently through get joy from. It's seen as a portion of computing. Machine picking up information of calculations build a adaptation bolstered design realities, alluded to as "preparing data", with a reason to form forecasts or choices while not being explicitly modified to achieve that. Machine finding out calculations unit of measurement utilized in an awfully wide frame of programs, comprising of e mail sifting and tablet innovative and perceiving, whereby it's miles extreme or infeasible to create standard calculations to perform the required obligations. [36] A set of machine almost to get it is closely related with method records, that creates a forte of building forecasts the work of pc frameworks; but not all gadget learning is connected arithmetic acing. The arranging at of numerical advancement offers strategies, arrange and application space names to the world of framework gaining data of. Information preparing might be a associated subject of take a look at, that concentrate on beta insights examination via unattended picking up information of. In its application all through undertaking issues, gadget learning in addition called prognosticative analytics.
Machine learning incorporates computers finding in any case they will perform obligations whereas not being explicitly modified to achieve that. It includes computers picking up information of from records prepared in arrange that they perform bound errands. for direct assignments apportioned to computers, it's conceivable to code calculations telling the gadget a few ways to execute all steps required to resolve the trouble to hand; at the computer's portion, no almost to get it is required.[33,34] For a part of progressed duties, it may be extreme for an individual's to physically deliver the required calculations. In work out, it may develop to be encourage viable to assist the machine broaden its personal set of rules, rather than having human software engineers indicate each required step. The subject of gadget acing utilizes different strategies to coach pc frameworks to achieve commitments whereby no completely wonderful set of rules is realistic.[46,47] In cases wherever sizable numbers of capability answers exist, one strategy is to name assortment of the proper answers as substantial. this may at that point be utilized as coaching insights for the versatile computer to strengthen the set of rules(s)[35,37] it makes utilize of to work out adjust arrangements. as an illustration, to coach a instrument for the wander of virtual man or young lady acknowledgment, the MNIST dataset of composed digits has more often than not been utilized.
Machine Learning Approaches
Machine learning approaches square measure historically divided into 3 broad classes, looking on the character of the "signal" or "feedback" obtainable to the educational system:
Supervised learning: The tablet is outfitted illustration inputs and their most well-liked yields, given by an "educator", and thus the objective is to inquire about a far reaching run the show that maps inputs to yields. Supervised gaining data of calculations area unit apt the utilization of classified illustrations, that grasp relate degree input whereby the favored yield is known. As a case, a touch of pack may have information variables classified either “F” (fizzled) or “R”
(runs). The instructive set of rules gets a tough and fast of inputs adjacent to the comparing rectify yields, and so the algorithmic program learns by means of assessing its real yield with redress yields to urge mistakes. It at that point adjusts the adaptation thus. Through methodologies like course, relapse, forecast and slope boosting, administered around to get a handle on makes utilize of plans to anticipating the values of the name on extra untagged insights. Administered finding out is regularly used in applications amid which memorable information predicts likely future events. as an illustration, it'll expect whereas Master card exchanges unit all told likelihood to be beguiling or that scope benefactor is likely to report a claim.
Unsupervised learning: No names unit of measurement given to the picking up data of run the show, flight it on its individual to discover shape in it enter. Unattended picking up data of are frequently an objective in itself (finding covered up designs in records) or the way inside the course of Relate in Nursing wrap up (work going to know). Unattended picking up data of is utilized against measurements that has no ancient labels. The framework isn't tutored the
"correct reply." The run the show got to find what is being prove. The purposeful is to discover the information and discover out a few structures among. Unattended acing works pleasantly on value-based information. For occasion, it may get it sections of customers with comparable qualities Joined together Countries office will at that point be limited similarly in promoting campaigns. Or it's prepared to take note the rule qualities that isolated emptor portions from each totally distinctive.[18,22,30] Elegant methods envelop self-organizing maps, nearest-
neighbor mapping, okay-manner cluster and solitary esteem deterioration. These calculations are moreover usual area matter substance themes, support contraptions and get it insights exceptions.
Semi supervised learning: It is utilized for steady programs as directed picking up data of. in any case its employments, each classified and untagged truths for coaching – ordinarily a little amount of classified data with a colossal sum of untagged measurements (since untagged actualities is way less extravagant and takes less organize construct up">to construct). Up kind this sort this kind} of pointing to recognize is additionally utilized with methods like sort, relapse and forecast. Semi directed learning is supportive while the worth related to labeling is fair as well intemperate to allow for a really labelled tutoring strategy. Early tests of this incorporate choosing a personality confront on a webcam.
Reinforcement learning: A pc application inter atomic with an energetic setting whereby it ought to perform a certain point (in conjunction with utilizing a car or taking portion in a diversion against relate rival). Since it navigates its disadvantage region, the program is outfitted input it's practically equivalent to rewards, that it tries to boost. Fortification acing is commonly utilized for manufactured insights, betting and navigation. With support advancing to recognize, the set of rules finds through trial and mistake that developments abdicate the finest rewards. This sort of acing has three favored added substances: the specialist (the learner or call producer), the environment (the entireness the operator inter atomic with) and activities (what the specialist will do). The objective is for the specialist to settle on developments that maximize the anticipated remunerate over a given amount of your time. The specialist can reach the point parts quicker with the assistance of taking after amarvelous arrangement. That the point in support advancing to recognize is to examine the fine approach.
Conclusion
Confront cover discovery is additionally a troublesome assignment. It has been accepting a parcel of a part of consideration all through this time since of the spreading of crown infection sickness. Thus a few nations taking after the run the show like “No passage whereas not mask”. Veil discovery is exceptionally crucial issue in security reason and Covid-19 obstructions. Among the case of restorative field, veil diminishes potential introduction chance from an tainted individual whether or not or not they require side effects or not. Cover location is utilized in Airplane terminals, Clinics, Workplaces and instructional exercise Divisions etc.
in this way cover discovery is gotten to be an amazingly significant and troublesome issue. The confront acknowledgment whereas not veil troublesome} be that as it may confront acknowledgment with cover is vital one as a result of highlight extraction of cloaked confront is exceptionally complicated than conventional confront. That's changed confront choices like nose, mouth and chin are truant among the cloaked confront. In restorative field, veil decreases potential exposure‟s chance from an contaminated individual whether or not or not they require indications or not. The manufactured cleverly (AI) and machine learning (ML) are created various models for cover location. All through this study, specified with respect to various methodologies are utilized for facial cover location. By victimization machine learning algorithmic rule, we are getting to essentially take note the facial cover. And conjointly offer tall power in veil location at various genuine time situations.
References
[1] V. Chamola, V. Hassija, V. Gupta, and M. Guizani, „„A comprehensive review of the COVID-19 pandemic and the role of IoT, drones, AI, blockchain, and 5G in managing itsimpact,‟‟ IEEE Access, vol. 8, pp. 90225–90265, 2020.
[2] A. A. Hussain, O. Bouachir, F. Al-Turjman, and M. Aloqaily, „„AI techniques for COVID-19,‟‟ IEEE Access, vol. 8, pp. 128776–128795,2020.
[3] Q.-V. Pham, D. C. Nguyen, T. Huynh-The, W.-J.Hwang, and P. N. Pathirana,
„„Artificial intelligence (AI) and big data for coronavirus (COVID-19) pandemic: A survey on the state-of-the-arts,‟‟ IEEE Access, vol. 8, pp. 130820–130839,2020.
[4] M. Loey, G. Manogaran, M. H. N. Taha, and N. E. M. Khalifa, „„A hybrid deep transfer learning model with machine learning methods for face mask detection in the era of the COVID-19 pandemic,‟‟ Meas., J. Int. Meas. Confederation, vol. 167, May 2020, Art. no.108288.
[5] Murugesan, M., Thilagamani, S. ,” Efficient anomaly detection in surveillance videos based on multi layer perception recurrent neural network”, Journal of Microprocessors and
Microsystems, Volume 79, Issue November 2020,
https://doi.org/10.1016/j.micpro.2020.103303
[6] A. Waheed, M. Goyal, D. Gupta, A. Khanna, F. Al-Turjman, and P. R. Pinheiro,
„„CovidGAN: Data augmentation using auxiliary classifier GAN for improved COVID- 19 detection,‟‟ IEEE Access, vol. 8, pp. 91916–91923,2020.
[7] Thilagamani, S., Nandhakumar, C. .” Implementing green revolution for organic plant forming using KNN-classification technique”, International Journal of Advanced Science and Technology, Volume 29 , Isuue 7S, pp. 1707–1712
[8] X. Ouyang, J. Huo, L. Xia, „„Dual-sampling attention network for diagnosis of COVID-19 from community acquired pneumonia,‟‟ IEEE Trans. Med. Imag., vol. 39, no. 8, pp. 2595–2605, Aug.2020.
[9] Thilagamani, S., Shanti, N.,” Gaussian and gabor filter approach for object segmentation”, Journal of Computing and Information Science in Engineering, 2014, 14(2), 021006, https://doi.org/10.1115/1.4026458
[10] J. Li, J. Wang, and F. Ullah, „„An end-to-end task-simplified and anchorguided deep learning framework for image-based head pose estimation,‟‟ IEEE Access, vol. 8, pp.
42458–42468,2020.
[11] Rhagini, A., Thilagamani, S. ,”Women defence system for detecting interpersonal crimes”, International Journal of Advanced Science and Technology, 2020, Volume 29,Issue7S, pp. 1669–1675
[12] N. Otberdout, A. Kacem, M. Daoudi, L. Ballihi, and S. Berretti, „„Automatic analysis of facial expressions based on deep covariance trajectories,‟‟ IEEETrans.
Neural Netw. Learn. Syst., vol. 31, no. 10, pp. 3892–3905, Oct.2020.
[13] K.Deepa, S.Thilagamani, “Segmentation Techniques for Overlapped Latent Fingerprint Matching”, International Journal of Innovative Technology and Exploring Engineering (IJITEE), ISSN: 2278-3075, Volume-8 Issue-12, October 2019. DOI: 10.35940/ijitee.L2863.1081219
[14] Deepa, K., Kokila, M., Nandhini, A., Pavethra, A., Umadevi, M. “Rainfall prediction using CNN”, International Journal of Advanced Science and
Technology, 2020, 29(7 Special Issue), pp. 1623–1627.
http://sersc.org/journals/index.php/IJAST/article/view/10849.
[15] J. Wang, F. Ullah, Y. Cai, and J. Li, „„Non-stationary representation for continuity aware head pose estimation via deep neural decision trees,‟‟ IEEE Access, vol. 7, pp.
181947–181958, 2019.
[16] Santhi, P., Priyanka, T.,Smart India agricultural information reterival system, International Journal of Advanced Science and Technology, 2020, 29(7 Special Issue), pp.
1169–1175.
[17] D. Lee, M.-H. Yang, and S. Oh, „„Head and body orientation estimation using convolutional random projection forests,‟‟ IEEE Trans. Pattern Anal. Mach. Intell., vol.
41, no. 1, pp. 107–120, Jan.2019.
[18] Santhi, P., Lavanya, S., Prediction of diabetes using neural networks, International Journal of Advanced Science and Technology, 2020, 29(7 Special Issue), pp. 1160–1168.
[19] B. Ahn, J. Park, and I. S. Kweon, „„Real-time head orientation from a monocular camera using deep neural network,‟‟ in Proc. Asian Conf. Comput. Vis., in Lecture Notes in Computer Science, vol. 9005, 2015, pp.82–96.
[20] Santhi, P., Mahalakshmi, G., Classification of magnetic resonance images using eight directions gray level co-occurrence matrix (8dglcm) based feature extraction, International Journal of Engineering and Advanced Technology, 2019, 8(4), pp. 839–
846.
[21] M. Raza, Z. Chen, S.-U.Rehman, P. Wang, and P. Bao, „„Appearance based pedestrians‟ head pose and body orientation estimation using deep learning,‟‟
Neurocomputing, vol. 272, pp. 647–659, Jan.2018.
[22] Y. Lu, S. Yi, N. Hou, J. Zhu, and T. Ma, „„Deep neural networks for headpose classification,‟‟ in Proc. World Congr. Intell. Control Autom., Sep. 2016, pp. 2787–2790.
[23] J. Bao and M. Ye, „„Head pose estimation based on robust convolutional neural network,‟‟ Cybern. Inf. Technol., vol. 16, no. 6, pp. 133–145, Dec.2016.
[24] N. Ruiz, E. Chong, and J. M. Rehg, „„Fine-grained head pose estimation without keypoints,‟‟ in Proc. IEEE/CVF Conf. Comput. Vis. Pattern Recognit. Workshops (CVPRW), Jun. 2018, pp.2155–2164.
[25] Vijayakumar, P, Pandiaraja, P, Balamurugan, B & Karuppiah, M 2019, „A Novel Performance enhancing Task Scheduling Algorithm for Cloud based E-Health Environment‟, International Journal of E-Health and Medical Communications , Vol 10,Issue 2,pp 102-117
[26] P. Li, Y. Li, and L. Tan, „„Transfer useful knowledge for headpose estimation from low resolution images,‟‟ Multimedia Tools Appl., vol. 75, no. 15, pp. 9395–9408, Aug.
2016.
[27] Y. Liu, P. Lasang, S. Pranata, S. Shen, and W. Zhang, „„Driver pose estimation using recurrent lightweight network and virtual data augmented transfer learning,‟‟IEEE Trans.
Intell. Transp. Syst., vol. 20, no. 10, pp. 3818–3831, Oct. 2019.
[28] P. Pandiaraja, N Deepa 2019 ,” A Novel Data Privacy-Preserving Protocol for Multi-data Users by using genetic algorithm” , Journal of Soft Computing , Springer , Volume 23 ,Issue 18, Pages 8539-8553.
[29] P. Sreekanth, U. Kulkarni, S. Shetty, and M. S. M., „„Head pose estimationusing transfer learning,‟‟ in Proc. Int. Conf. Recent Trends Advance Comput. (ICRTAC), Sep.
2018, pp. 73–79.
[30] N Deepa , P. Pandiaraja, 2020 ,” Hybrid Context Aware Recommendation System for E-Health Care by merkle hash tree from cloud using evolutionary algorithm” , Journal of Soft Computing , Springer , Volume 24 ,Issue 10, Pages 7149–7161.
[31] R. K. Samala, H.-P. Chan, L. Hadjiiski, M. A. Helvie, C. D. Richter, and K. H. Cha,
„„Breast cancer diagnosis in digital breast tomosynthesis: Effects of training sample size on multi-stage transfer learning using deep neural nets,‟‟ IEEE Trans. Med. Imag., vol. 38, no. 3, pp. 686–696, Mar.2019.
[32] N Deepa , P. Pandiaraja, 2020 , “ E health care data privacy preserving efficient file retrieval from the cloud service provider using attribute based file encryption “, Journal of Ambient Intelligence and Humanized Computing , Springer , https://doi.org/10.1007/s12652-020-01911-5.
[33] Z. Wang, L. Du, J. Mao, B. Liu, and D. Yang, „„SAR target detection based on SSD with data augmentation and transfer learning,‟‟ IEEE Geosci. Remote Sens. Lett., vol. 16, no. 1, pp. 150–154, Jan.2019.
[34] K Sumathi, P Pandiaraja 2019,” Dynamic alternate buffer switching and congestion control in wireless multimedia sensor networks” , Journal of Peer-to-Peer Networking and
Applications , Springer , Volume 13,Issue 6,Pages 2001-2010.
[35] D. Malmgren-Hansen, A. Kusk, J. Dall, A. A. Nielsen, R. Engholm, and H.
Skriver, „„Improving SAR automatic target recognition models with transfer learning from simulated data,‟‟ IEEE Geosci. Remote Sens. Lett., vol. 14, no. 9, pp. 1484–
1488, Sep. 2017.
[36] Sumathi, K., Naveena, K., Prashanth, P., Revanthkumar, S., Srikeerthanaa, A.P.,E-health based patient surveilling device, International Journal of Emerging Trends in Engineering Research, 2020, 8(3), pp. 792–796
[37] T. Shermin, S. Wei Teng, M. Murshed, G. Lu, F. Sohel, and M. Paul, „„Enhanced transfer learning with ImageNet trained classification layer,‟‟ 2019, arXiv:1903.10150.
[Online]. Available:http://arxiv.org/abs/1903.10150
[38] B. Zhao, B. Huang, and Y. Zhong, „„Transfer learning with fully pretrained deep convolution networks for land-use classification,‟‟ IEEE Geosci. Remote Sens. Lett., vol.
14, no. 9, pp. 1436–1440, Sep.2017.
[39] Shankar, A., Pandiaraja, P., Sumathi, K., Stephan, T., Sharma, P. ,” Privacy preserving E-voting cloud system based on ID based encryption ” Journal of Peer-to-Peer Networking and Applications , Springer , https://doi.org/10.1007/s12083-020-00977-4.
[40] H. Yao, S. Zhang, Y. Zhang, J. Li, and Q. Tian, „„Coarse-to-fine description for fine-grained visual categorization,‟‟ IEEE Trans. Image Process., vol. 25, no. 10, pp.
4858–4872, Oct.2016.
[41] L. Zheng, Y. Zhao, S. Wang, J. Wang, and Q. Tian, „„Good practice in CNN feature transfer,‟‟ 2016, arXiv:1604.00133. [Online]. Available:http://arxiv.org/abs/1604.00133
[42] Y. LeCun, L. Bottou, Y. Bengio, and P. Haffner, „„Gradient-based learning applied to document recognition,‟‟ Proc. IEEE, vol. 86, no. 11, pp. 2278–2323, Nov.1998.
[43] A. Krizhevsky, I. Sutskever, and G. E. Hinton, „„ImageNet classification with deep convolutional neural networks,‟‟ Commun. ACM, vol. 60, no. 6, pp. 84–90, May2017.
[44] S. Jialin Pan and Q. Yang, „„A survey on transfer learning,‟‟ IEEE Trans. Knowl.
Data Eng., vol. 22, no. 10, pp. 1345–1359, Oct.2010.
[45] J. Deng, W. Dong, R. Socher, L.-J. Li, K. Li, and L. Fei-Fei, „„ImageNet: A large- scale hierarchical image database,‟‟ in Proc. IEEE Conf. Comput. Vis. Pattern Recognit., Jun. 2009, pp. 248–255.
[46] S. Kornblith, J. Shlens, and Q. V. Le, „„Do better ImageNet models transfer better?‟‟
in Proc. IEEE/CVF Conf. Comput. Vis. Pattern Recognit. (CVPR), Jun. 2019, pp. 2656–
2666.
[47] W. Gao, B. Cao, S. Shan, X. Chen, D. Zhou, X. Zhang, and D. Zhao, „„The CAS-
PEAL large-scale chinese face database and baseline evaluations,‟‟ IEEE Trans. Syst., Man, Cybern. A, Syst. Humans, vol. 38, no. 1, pp. 149–161, Jan.2008.
[48] S. Mehdi Iranmanesh and N. M. Nasrabadi, “Attribute-guided deep polarimetric thermal-to-visible face recognition,” 2019, arXiv:1907.11980. [Online].
Available:http://arxiv.org/abs/1907.11980
[49] V. Talreja, F. Taherkhani, M. C Valenti, and N. M. Nasrabadi, “Attributeguided coupled GAN for cross-resolution face recognition,” 2019, arXiv:1908.01790. [Online].
Available:http://arxiv.org/abs/1908.01790
[50] X. Zhu et al., “Large-scale bisample learning on ID versus spot face recognition,”
Int. J. Comput. Vis., vol. 127, nos. 6–7, pp. 684–700, Jun.2019.
