INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue VII, July 2025

www.ijltemas.in Page 1101

An Intelligent System for Face Mask Recognition and Non-Contact

Temperature Detection Using Deep Learning

Dr S Arun, Mr. A. Kuppuswamy

Associate Professor, EEE Department, P A College of Engineering and Technology, Pollachi, Tamilnadu

DOI: https://doi.org/10.51583/IJLTEMAS.2025.1407000132

Abstract— The increasing prevalence of airborne infectious diseases has necessitated the development of intelligent, automated

screening systems to ensure public safety in high-risk environments. This paper proposes a deep learning and neural network-based

approach for integrated face mask detection and non-contact temperature identification. The system is designed to operate in real-

time and is suitable for deployment in public spaces such as transportation terminals, corporate entry points, and commercial

facilities. To enhance identification accuracy under mask-wearing conditions, the proposed framework employs a masked facial

recognition technique that utilizes convolutional neural networks (CNNs) to analyze visible facial features. In parallel, a contactless

infrared temperature sensor enables real-time thermal screening without human intervention. The system architecture is supported

by a lightweight, reliable IoT communication protocol for efficient data transmission to a cloud-based platform, enabling remote

monitoring through web and mobile applications. Experimental validation demonstrates that the proposed system achieves high

accuracy in both mask detection and temperature measurement, outperforming conventional manual methods in terms of speed,

reliability, and scalability. The collected data can be further utilized for health analytics and decision support by relevant authorities.

This work contributes a cost-effective, scalable solution for enhancing public health safety through automation and intelligent

sensing.

Keywords— Face mask detection, temperature measurement, deep learning with neural networks, IoT

I. Introduction

The recent outbreak of a highly contagious respiratory illness, which originated in Wuhan, China, rapidly spread across numerous

countries, including India—the second most populous nation globally. Due to its dense population, India faced significant challenges

in controlling the transmission of this viral infection. Among the most effective preventive measures are the use of face masks and

temperature screening systems, which have shown promising results in minimizing the rate of transmission. Common symptoms

associated with such respiratory infections include fever, sore throat, fatigue, loss of taste and smell, and nasal congestion, as noted

in 1. Transmission primarily occurs through respiratory droplets and indirect contact via contaminated surfaces, with an incubation

period that can extend up to 10 to 14 days in severe cases.

To mitigate the spread of the disease, authorities implemented a range of public health protocols, including social distancing,

mandatory mask usage in enclosed environments, quarantine measures, travel restrictions, self-isolation guidelines, and the

suspension of mass gatherings and events, as detailed in 2. These precautionary measures have significantly impacted daily life,

affecting work environments, social engagements, sports activities, and entertainment sectors. Individuals exhibiting elevated body

temperature should be restricted from accessing public spaces to prevent further spread, while mask compliance should be strictly

enforced. Accordingly, temperature checks and face mask verification have become mandatory at various public entry points,

including airports, toll plazas, workplaces, shopping centers, and healthcare facilities, as outlined in 3. To address these requirements,

a smart entry system has been developed that integrates automated face mask detection with contactless body temperature

measurement. This system employs advanced technologies to enable a safe and efficient access control mechanism, enhancing public

safety in environments prone to high human interaction.

The remainder of this paper is organized as follows. Section II provides a comprehensive review of related work and methodologies

presented in the existing literature. Section III outlines the structural organization of the proposed system. Section IV offers an

overview of the convolutional neural network (CNN) methodology employed for face mask detection and temperature analysis.

Section V presents the system architecture and operational workflow of the proposed solution. Section VI discusses the experimental

results and performance evaluation. Finally, Section VII concludes the paper and highlights potential directions for future research.

II. Literature Review

Before you begin to format your paper, first write and save the content as a separate text file. Complete all content and organizational

editing before formatting. Please note sections A-D below for more information on proofreading, spelling, and grammar.

Face mask detection 7 is detection by using a deep learning algorithm to identify patterns. A prototype system for capturing online

attendance records was 8, and this system is a web-based Python program. Users can access the system interface from any browser,

regardless of terminal. The performance of the pre-trained CNN with multi-class support vector machine (SVM) classifier and the

performance of transfer learning utilizing the AlexNet model to do classification are investigated 9. The max pooling layer in CNN

reduces the resolution of the picture 10 patches while making the model more resilient than another standard feature extraction

technique known as local multiple pattern. The distance ratio between the classes has been maximized as a result of CNN optimization

in the linear collaborative discriminant regression classification. To extract the characteristics of the numerous convolution layers of

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS) 
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue VII, July 2025 
www.ijltemas.in                                                                                                                                                 Page  1102 
 
 
 
the low-resolution thermal infrared images, a pre-trained CNN is used 11 and utilizes a sample of 1500 scaled thermal photos with a 
resolution of 181 x 161 pixels, and its effectiveness is measured with a variety of moods, positions, and lighting circumstances. 
System Structure 
Figure 1 shows the structure of the proposed face mask and temperature detection system. A webcam's real-time video stream of 
human features is fed into the system 12 for mask recognition. Each frame of the video stream is subjected to various picture 
processing methods. A mask detector model was created based on a deep learning method known as Very Deep  Convolutional 
Networks for Large-Scale Image Recognition, and it actually detects masks. (VGG-16) 13. The model identifies mask features on 
each RoI to determine whether or not a mask exists. Finally, the GUI displays a label with the likelihood of the classification outcome. 
As a result, temperature data received from a serial port is fed into another thread for processing.  
The GUI displays temperature measurements, and if the temperature exceeds a predefined threshold of 37
o
C, staffs are notified that 
entry is not permitted. The block diagram depicts the overall system's workflow. The temperature sensor receives input from the 
person and sends 14-15 the data to the processor, which compares the temperature with the fixed range. If the temperature is within 
the fixed range, the processor sends the signal to the GUI, which scans the face and compares the input data with the library available 
in the           GUI module. If there is a mask on the human face, GUI sends the signal to the processor. 
 
Fig. 1.  The Structure of the suggested system 
Convolutional Neural Network  
A type of artificial neural network known as a convolutional neural network is used mostly for image processing, classification, 
segmentation, and other autocorrelated data, as shown in figure 2. It comprises one or more convolutional layers. A machine 
learning-based artificial neural network called deep learning can recognize things in images by gradually extracting attributes from 
the data as it moves up the layers. As depicted in the graphic, we must train the CNN with human faces in order to recognize faces 
in  images  16-17.  Utilizing  CNNs  has  the  advantage  of  allowing  for  the  internal  development  of  a  two-dimensional  image 
representation. This enables the model to understand the size and positioning of faces in an image. Pre-trained CNN models include 
18-19  AlexNet and GoogleNet. Nine layers make up the network that is being used here. These nine layers include one softmax 
regression layer 20, two fully connected layers, two pooling layers, and three convolution layers. Following the feature extraction 
process, two fully linked layers and a softmax classifier with potent non-linear classification capabilities are used as the final two 
layers. 
Overview of The Proposed Model 
The circuit setup as shown in figure 3 uploaded the deep learning code to the Arduino Uno board as in 21; the code will automatically 
upload because the ESP32-CAM is connected to the arduino uno board. By providing the Wi-Fi SSID and password in the arduino 
code, the ESP32-CAM will be connected to the appropriate Wi-Fi support after uploading the code. On the laptop, the same Wi-Fi 
support should be installed. The serial monitor of the arduino program generates the ESP32-CAM's IP address. The ESP32 CAM 
is now turned on and broadcasting live video as in 22, which can be seen on the laptop.  
Temperature Detection: The MLX90614 is a contactless infrared temperature sensor that can be used to assess the temperature of a 
specific object for the detection of body temperature, as in 23. The MLX90614 sensor measures the temperature of a specific object 
using infrared rays without any physical contact and communicates with the microcontroller via the I2C protocol. There are fixed 
boundaries in the system. If it exceeds the cut-off point, the buzzer will sound. 

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue VII, July 2025

www.ijltemas.in Page 1103

Fig. 2. Structure of deep learning with neural networks

Face Mask Detection: After checking the temperature, it sends a signal to the ESP32 to see if the human is wearing a mask. The

ESP32 will do this by taking a picture of the person and comparing it to a picture stored in a library or data set, as in 24. After the

comparison, it sends the current status of the human face to the arduino. The deep learning algorithm will evaluate images by segment;

once the evaluation is done, a signal is sent to the arduino controller.

Door Control: The door will open based on the two circumstances listed above. If either condition is not met, the arduino will not

send the signal to the motor to open the door. If both the conditions are met, the arduino will send a signal to the motor to open the

door, as in 25.

Fig. 3. Circuit diagram of the proposed system

Training and Evaluation

Based on a publicly available model, this study employed a pre-trained network using transfer learning. It has 16 prediction layers

with a decreasing feature-map size. Twenty percent of the annotated photos from the training data set were designated as a test set

prior to the network's training. After the network was trained, this evaluation set was used to test it in scenarios that it had not been

exposed to before.

Fig. 4. Person with proper mask

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue VII, July 2025

www.ijltemas.in Page 1104

Figure 4 illustrates the output for a person wearing a properly positioned mask. The system detects the mask with an accuracy of

97%.

Fig. 5. Person without a mask

Figure 5 displays the detection result for a person who is not wearing a mask. The face detection system successfully identifies the

absence of a mask and highlights the individual with a red bounding box. The confidence level for this detection is 99.21%, indicating

a high degree of accuracy in recognizing non-compliance with mask-wearing protocols.

Fig. 6. Person with improper mask

Figure 6 presents the detection result for a person wearing a mask incorrectly, where the mask is placed below the nose and mouth,

failing to provide proper coverage. The face detection system correctly identifies this as a case of non-compliance and highlights the

face with a red bounding box. The no-mask detection accuracy for this scenario is 69%, reflecting the system’s ability to detect

improper mask usage, though with relatively lower confidence compared to fully non-masked cases.

Fig. 7. ROC curve

Figure 7 presents a benchmark analysis using receiver operating characteristic (ROC) principles. The graph illustrates detection

performance across various facial regions, including right cheek, forehead, eyes, and full face coverage. The true positive (TP) and

false positive (FP) rates are affected by classification thresholds; reducing the threshold increases both genuine and false positives.

In this work, deep learning algorithms are employed for target recognition, image segmentation, and image classification using

different datasets: the ESP32 dataset for segmentation, ImageNet for object detection, and a custom dataset for classification. The

performance of each approach is evaluated objectively, with the proposed method consistently outperforming existing methods. The

achieved accuracy rates are 86% for target recognition, 88.3% for image segmentation, and 90.1% for image classification,

100

Percentage

Proposed

Method

[5]

[6]

[7]

[8]

[9]

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS) 
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue VII, July 2025 
www.ijltemas.in                                                                                                                                                 Page  1105 
 
 
 
demonstrating the robustness of the proposed approach. The quality of the mask detection system is evidenced through a range of 
scenarios. The system accurately identifies properly masked individuals (97% accuracy), unmasked individuals (99.21% accuracy), 
and those wearing masks incorrectly (69% accuracy). These results showcase the system’s sensitivity to real-world variations in mask 
usage, highlighting its effectiveness in both strict and ambiguous conditions. Such high detection accuracy, especially in challenging 
cases like partial masking, validates the reliability and practical applicability of the developed method for real-time compliance 
monitoring. 
III. Conclusion 
This work presents an intelligent, automated face mask and temperature detection system that integrates machine learning techniques 
with image processing for real-time public health screening. By employing CNNs for feature extraction, the system effectively 
enhances the accuracy of facial recognition, even under mask-wearing conditions. The dual-stage process comprising temperature 
validation  followed  by  mask  detection  ensures  that  access  control  is  both  secure  and  efficient,  without  the  need  for  human 
intervention. Experimental results demonstrate that the proposed approach significantly improves target recognition, classification, 
and segmentation performance compared to conventional methods. The integration of machine learning and image processing in this 
context highlights the growing relevance of artificial intelligence in public safety applications. As machine learning continues to 
evolve, its role in image-based recognition systems is expected to expand across various sectors. The findings of this study contribute 
to the ongoing development of intelligent monitoring technologies and provide a benchmark for future research in automated health 
screening and AI-assisted surveillance systems. 
References 
1.  Tanu Singhal, “A Review of Coronavirus Disease- 2019”, The Indian Journal of Pediatrics, Vol. 84, Is.no 4,  pg. 281-286, 
March 2020. 
2.  Ankit Viramgami , Avinash Pagdhune , Kamalesh Sarkar , Rakesh Balachandar, “Occupational Health and Safety Practices 
at Workplace during COVID-19 Pandemic” Journal of Comprehensive Health, Vol. 8, I.2, July-2020. 
3.  Mohammad Marufur Rahman, Md. Motaleb Hossen Manik, Md. Milon Islam, Saifuddin Mahmud, Jong- Hoon Kim. “An 
Automated  System  to  Limit  the  COVID-19  Using  Facial  Mask  Detection  in  Smart  City  Network”,  Electronics  and 
Mechatronics Conference (IEMTRONICS), 2020 IEEE International IOT. 
4.  Gorhekar, R. Shah, C. Shah, V. Bide, P.J. “A Survey on Covid-19 Face-Mask Detection Techniques”, In Proceedings of 
the  IEEE  4th  International  Conference  on  Computing,  Power  and  Communication  Technologies  (GUCON),  Kuala 
Lumpur, Malaysia, Institute of Electrical and Electronics Engineers: Piscataway, NJ, USA, pp. 1–6, 2018. 
5.  Mahaveer Penna, Shivashankar, B Arjun, K R Goutham, Lohith N Madhaw, Kumar G Sanjay. “Smart fleet monitoring 
system  using  Internet  of  Things  (IoT)”,  Recent  Trends  in  Electronics  Information  &  Communication  Technology 
(RTEICT) 2nd IEEE International Conference, pp. 1232-1236. 2018. 
6.  Gang Jin, Xiangyu Zhang, Wenqiang Fan, Yunxue Liu and Pengfei He. “Design of non-contact Infra-red thermometer 
based on the sensor of MLX90614”, The Open Automation and Control Systems Journal, Vol.7, pp. 8-20, 2015. 
7.  Sharma, A., Gautam, R. & Singh, J. “Deep learning for face mask detection: a survey”, Multimed Tools Appl, Vol. 82, 
Pg. 34321-34361, 2023. 
8.  Kamil, M.H.M.,  Zaini, N., Mazalan, L. et  al.,  “Online  attendance  system  based  on  facial  recognition  with  face  mask 
detection”. Multimed Tools Appl, Vol. 82, Pg. 34437-34457, 2023. 
9.  Almabdy  S,  Elrefaei  L.  “Deep  Convolutional  Neural  Network-Based  Approaches  for  Face  Recognition”. Applied 
Sciences, Vol. 9 Is. No. 20, 2019 
10.  Sangamesh Hosgurmath , Viswanatha Vanjre Mallappa, Nagaraj B. Patil, Vishwanath Petli, “A face recognition system 
using  convolutional  feature  extraction  with  linear  collaborative  discriminant  regression  classification”,  International 
Journal of Electrical and Computer Engineering (IJECE), Vol. 12, No. 2, pp. 1468~1476, April 2022. 
11.  Mohamed Sayed, Faris Baker, “Thermal Face Authentication with Convolutional Neural Network”, Journal of Computer 
Science, Vol.14, Is.no.12, pp. 1627-1637, 2018. 
12.  Elliot Mbunge, Sakhile Simelane, Stephen G Fashoto, Boluwaji Akinnuwesi,  Andile S  Metfula,. “Application of deep 
learning and machine learning models to detect COVID-19 face masks”, A review, Sustainable Operations and Computers, 
Vol. 2, ,Pages 235-245, 2021. 
13.  Simonyan, Karen, Zisserman, Andrew, “Very Deep Convolutional Networks for Large-Scale Image Recognition”, 3rd 
International Conference on Learning Representations, ICLR 2015, San Diego, CA, USA, May 7-9. 2014. 
14.  Jash Shah, Heth  Gala, Kevin Pattni  “Arduino  Based  Temperature,  Mask  Wearing  and  Social  Distance  Detection for 
COVID-19” In book: Internet of Things and Connected Technologies, vol.15, pp.148-161, January 2022. 
15.  Shyam, D.; Kot, A. Athalye, C. “Abandoned Object Detection Using Pixel-Based Finite State Machine and Single Shot 
Multibox Detector”, In Proceedings of the IEEE International Conference on Multimedia and Expo (ICME), San Diego, 
CA, USA; Institute of Electrical and Electronics Engineers: Piscataway, NJ, USA, pp. 1-6. 2018. 
16.  Prof. S. S. Bhong, Tanmay Hajare, Shreyank Bansod, Amitesh Sakharwade, Harshit less. (2022) ‘Face Mask and Social 
Distancing Detection System’, International Journal of Advanced Research in Science, Communication and Technology. 
17.  R. Thukral and M. Mishra, "An IoT Based Intelligent Screening System for Preventing COVID 19 Spread in Premises," 
2021 4th International Conference on Computing and Communications Technologies (ICCCT), pp. 175-180, 2021 

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS) 
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue VII, July 2025 
www.ijltemas.in                                                                                                                                                 Page  1106 
 
 
 
18.  Aneesa M P , Saabina N , Meera K, 2022, “Face Recognition using CNN: A Systematic Review”, International Journal 
Of Engineering Research & Technology,  Vol. 11, Is.No. 06, June 2022. 
19.  Arun,  S.,  &  Manigandan,  T.  (2021).  Design  of  ACO  based  PID  controller  for  zeta  converter  using  reduced  order 
methodology. Microprocessors and Microsystems, 81, 103629. 
20.  K. Yan, S. Huang, Y. Song, W. Liu and N. Fan, "Face recognition based on convolution neural network," 2017 36th 
Chinese Control Conference (CCC), Dalian, China, pp. 4077-4081, 2017. 
21.  Leo Louis “Working Principle of Arduino and Using it as a Tool for Study and Research” International Journal of Control, 
Automation, Communication and Systems (IJCACS), April 2018. 
22.  Dilip  Prathapagiri,  Eethamakula  Kosalendra  “Wi  -  Fi  Door  Lock  System  Using  ESP32  CAM  Based  on  IoT”,  The 
International journal of analytical and experimental modal analysis, July 2021. 
23.  Gonçalo Marques, Rui Pitarma “Non-contact Infrared Temperature Acquisition System based on Internet  of Things for 
Laboratory Activities Monitoring” The 9th international conference on sustainable energy information technology(SEIT) 
Procedia Computer Science 155, pp. 487-494, August 2019. 
24.  Ponlawat Khamlae, Kingkarn Sookhanaphibarn, Worawat Choensawat “An Application of Deep- Learning Techniques to 
Face Mask Detection During the COVID-19 Pandemic” Conference: 2021 IEEE 3rd Global Conference on Life Sciences 
and Technologies (LifeTech), March 2021. 
25.  Everistus Zeluwa Orji, Oleka Cv, Ujunwa Ifeoma Nduanya, “Arduino Based Door Automation System Using Ultrasonic 
Sensor and Servo Motor”, The Journal of Scientific and Engineering Research Vol.05, Is.No.04, pp.:341-349, April 2018. 