Volume 9, Issue 12, December – 2024 International Journal of Innovative Science and Research Technology

ISSN No:-2456-2165

Police Surveillance System for Missing Persons

Sumit Raj; Utkarsh Pratap Singh; Sumanth P N; Nayana C P
Department of Information Science
RNS Institute of Technology, Bengaluru

Abstract:- The rapid progress in facial recognition AI-enhanced learning systems bring a new level of relevance
technology has broadened its use across multiple sectors, to the educational process, enabling the identification of areas
with law enforcement and public safety being among the where learners may need additional support. AI-driven instant
primary areas of impact. This study presents an feedback mechanisms provide students with real-time
innovative framework for identifying criminals and performance insights, fostering a growth mindset and
missing persons by integrating two cutting-edge encouraging consistent progress. This personalized feedback
technologies: FaceNet and MTCNN. FaceNet, a deep is instrumental in helping learners recognize their strengths
learning-based model, produces high-dimensional facial and weaknesses, ultimately supporting a deeper grasp of the
embeddings that capture unique facial features subject. However, while AI offers promising advancements in
consistently across various conditions, while MTCNN education, several challenges must be overcome to fully
performs real-time face detection, isolating facial regions harness its potential. Concerns related to data privacy, the
accurately to improve identification precision. The accuracy and reliability of AI-generated content, and the need
combined application of FaceNet and MTCNN addresses for educator training in these technologies present significant
common challenges in facial identification, such as obstacles.
changes in lighting, pose, and expression, providing law
enforcement with a robust tool to expedite investigations II. LITERATURE SURVEY
and locate missing individuals. Through testing on diverse
datasets, this study assesses the system's effectiveness, A literature review is crucial as it examines prior
focusing on practical applicability and examining ethical analyses and studies conducted in the field of interest. It
concerns, privacy protections, and potential societal evaluates past findings, considering various aspects and the
impacts. This research contributes to the ongoing overall scope of the project. The primary purpose of a
discussion on using advanced technologies responsibly to literature review is to provide a thorough overview of the
enhance public safety and support law enforcement project's background, identifying weaknesses in the current
efforts. framework and highlighting unresolved issues. The topics
discussed not only offer insight into the project's background
Keywords:- Artificial Intelligence, E-Learning, Deep but also emphasize the challenges and limitations that
Learning, Real-time Face Detection, Adaptive Learning, inspired the project’s development and proposed solutions.
Natural Language Processing, Diverse Datasets, Learner
Engagement, Identification System, Data Privacy. Rajesh Kumar Tripathi-[1] presents a novel approach to
local feature extraction aimed at enhancing age-invariant face
I. INTRODUCTION recognition. It addresses the challenges posed by facial aging,
which can significantly impact recognition accuracy. The
The advancement of facial recognition technology has authors propose innovative techniques that improve the
had a profound impact across various fields, with law robustness of facial features against age variations. Their
enforcement and public safety being among the most affected. methodology is evaluated against existing methods,
As criminal activity continues and authorities face ongoing demonstrating superior performance in maintaining
challenges with missing person cases, there is a growing recognition accuracy across different age groups.
demand for sophisticated identification systems capable of
accurately matching faces across large databases. This V. Kaspate, et at. - [12] have proposed an AI-based
research proposes a unique method for identifying criminals framework for detecting missing people using CNN
and missing persons by integrating two state-of-the-art algorithms. The system integrates facial recognition, image,
technologies: FaceNet and MTCNN. FaceNet, a deep and video analysis, and automates data matching for quick
learning-based facial recognition model, is notable for its identification.
ability to produce high-dimensional facial embeddings that
capture distinct features of a face. This approach overcomes  It Employs Deep Learning to Process
traditional limitations by delivering a stable representation of A. Kawade, Y. Mahajan, H. Meherkhamb, A. Malvatkar
faces across different images and environmental conditions. - [15] have proposed a system for identifying missing persons
Complementing this, MTCNN (Multi-task Cascaded and criminals using image processing and facial recognition.
Convolutional Networks) is utilized for real-time face The system compares images from a database to web video
detection, effectively isolating facial regions from images to streams, automatically notifying police of matches. It
optimize the data for subsequent identification In addition, accounts for variations due to aging, lighting, and other

IJISRT24DEC699 www.ijisrt.com 547

Volume 9, Issue 12, December – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165

factors, streamlining the process of locating individuals including fatigue detection in drivers. The literature
across public spaces. demonstrates a clear progression toward more robust and
reliable systems that can handle real-time detection and
CHAP, Ramyadevi, K. Balasubramanian - [16] It tracking of individuals in various scenarios, contributing to
discusses the integration of historical criminal data and real- improved criminal and missing person identification
time CCTV feeds to enhance identification. Challenges capabilities.
include handling varying environmental conditions and
ensuring scalability. The study also focuses on optimizing W. K. Aljohani, R. A. Alshehri -[7] have introduced
recognition to support proactive policing. Suhail, a deep learning-based system for identifying missing
people. This Android-based system utilizes face recognition
TY JOUR , Kumar, K.Kranthi, Indira - [17] Researchers have techniques to minimize the time taken to locate missing
focused on addressing key challenges in face detection and persons by matching photos from a database. The system
recognition through various implementations, including the engages volunteers and government entities to improve the
integration of MTCNN (Multi-task Cascaded Convolutional identification process and reduce search time.
Networks) which has shown significant improvements in
handling variations in pose, illumination, and occlusion. S. Sambolek, M. Ivašić-Kos - [8] have proposed a
person detection system using drone imagery for search and
Shtwai Alsubai et al.-[2] introduces a novel age- rescue operations. They utilized Faster R-CNN with FPN as
invariant face recognition model that integrates Bald Eagle the backbone, applying transfer learning to improve detection
Search Optimization with deep transfer learning techniques. accuracy on datasets like VisDrone and a custom dataset
This approach effectively addresses the challenges of facial (SARD), significantly enhancing precision and recall in
aging by leveraging advanced optimization methods to challenging conditions.
enhance feature extraction. The authors demonstrate the
model's efficacy through extensive experiments, showcasing Akansha. D. Patil , Raj.H.Dubey- [9] The literature
improved recognition performance across various age groups. review highlights the use of Haar Cascade and AdaBoost
algorithms for real-time face detection, emphasizing
Jyothi S. Nayak, M. Indiramma-[3] have propose using challenges in detecting partially visible individuals in
PCA to create gender-specific eigen spaces, reducing the crowded scenes. It explores applications of machine vision
number of comparisons in face recognition. Validated on for identifying missing persons and assisting visually
ORL, Indian, and FG-NET databases, the method improves impaired individuals through voice alerts. The paper
accuracy and response time, especially for images with aging integrates OpenCV and Python to enhance detection
variations. efficiency.

Z. An, W. Deng - [4] have proposed an adaptive pose V. Shelke, T. Bangera - [10] have proposed
alignment method (APA). In the system they proposed they "Searchious," a system utilizing KNN-based face recognition
aimed to reduce intra-class variability due to pose differences, to locate missing individuals. It integrates an Android app for
increase the accuracy of face recognition, and demonstrated civilians and desktop software for police to enhance the
enhanced performance on IJB-A, IJB-C, and CPLFW datasets efficiency of tracing and registering new cases. The system
by aligning faces to optimal templates based on pose encodes facial points to improve detection accuracy and
estimation. facilitates quick alerts upon successful identification.

A. A M, A. G RAj - [5] have proposed an efficient Shakeel, M. Saad, Kin-Man Lam.- [11] The literature
tracking system for missing persons using AI. In the system review explores challenges in age-invariant face recognition
they proposed they aimed to utilize AI algorithms to analyze (AIFR), highlighting issues with aging variations affecting
data from various sources, improve face recognition accuracy shape and texture. Recent approaches include deep learning
using CNNs, and integrate predictive models to assist frameworks like CNNs, feature encoding, and canonical
authorities in locating missing individuals quickly and correlation analysis (CCA) to improve recognition accuracy.
efficiently. The review emphasizes the need for robust models that
account for both identity preservation and aging progression.
Roshin John , Basil Kuriakose-[6] It builds on machine
vision applications for identifying missing persons and Zhizhong Huang ,Graduate Student Member- [14]
assisting visually impaired individuals through voice alerts. explores developments in age-invariant face recognition
OpenCV and Python are used to develop the system, ensuring (AIFR) and face age synthesis (FAS). AIFR methods aim to
scalability and efficiency. The literature emphasizes the social disentangle identity-related features from age-related ones
impact of face recognition in public safety, human trafficking using techniques such as hidden factor analysis and CNN-
prevention, and enhanced surveillance. based frameworks. FAS methods use models like GANs and
conditional autoencoders to generate realistic age-progressed
Swapnil Joshi, Chaitanya Mhaske -[13] Studies have images, although balancing identity preservation remains
shown significant progress in face detection algorithms by challenging.
incorporating color space analysis with PCA verification,
while also addressing challenges in different environments

IJISRT24DEC699 www.ijisrt.com 548

Volume 9, Issue 12, December – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165

III. OUTCOME V. ADVANTAGES OF PROPOSED SYSTEM

 Enhanced Identification Accuracy:  High Accuracy in Identification:

By integrating FaceNet and MTCNN, the system The combination of MTCNN and FaceNet allows for
improves the accuracy of identifying criminals and missing precise facial recognition across different lighting, pose, and
persons, providing consistent facial recognition even with expression variations, reducing false matches and enhancing
variations in lighting, poses, and expressions. identification accuracy.

 Efficient Real-Time Detection:  Real-Time Processing

This real-time detection and identification system MTCNN enables quick face detection, allowing for real-
supports law enforcement in swiftly isolating and identifying time identification, which is crucial for timely investigations
individuals, which accelerates investigative processes. in law enforcement and locating missing persons.

 Ethical and Privacy-Focused Implementation:  Ethical and Privacy-Focused Design

The framework is designed with ethical considerations The system includes secure data handling, restricted
and privacy protections to address societal impacts, fostering access, and anonymization features to address privacy
transparency and building public trust in the technology’s concerns.
usage.
VI. METHODOLOGY
IV. PROPOSED SYSTEM
 Face Detection with MTCNN:
The proposed system combines MTCNN and FaceNet to Employ MTCNN to detect and isolate facial regions in
develop an effective, real-time facial recognition solution for images, providing reliable accuracy across varying lighting,
identifying criminals and missing individuals. Initially, angles, and expressions for consistent data input.
MTCNN detects and isolates facial regions within images,
ensuring accurate capture across different angles, lighting  Feature Extraction with FaceNet:
conditions, and expressions. FaceNet then generates unique Utilize FaceNet to create unique facial embeddings,
facial features by creating high-dimensional embeddings that generating high- dimensional vectors that capture each
remain consistent and reliable across varied conditions. These individual’s distinctive facial features for accurate
embeddings are matched against an existing database, using identification.
similarity scores to identify individuals of interest. For
practical reliability, the system is thoroughly evaluated on  Matching Against Database:
datasets to assess accuracy, speed, and false-positive rates. Compare the extracted facial embeddings with an
Ethical measures, such as secure data management, existing database of known individuals, calculating similarity
anonymization, and controlled access, are prioritized to scores to effectively identify or verify potential matches.
safeguard privacy and prevent misuse. This system offers law
enforcement a powerful, responsible tool for precise, real-  System Performance Evaluation:
time identification. Assess the system on diverse datasets, evaluating
metrics such as accuracy, speed, and error rates to confirm its
effectiveness and optimize for real-world application.

 Ethical and Privacy Safeguards:

Incorporate data privacy protocols, including secure
storage, restricted access, and anonymization, to ensure
ethical use and compliance with privacy standards.

IJISRT24DEC699 www.ijisrt.com 549

Volume 9, Issue 12, December – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165

VII. SYSTEM ARCHITECTURE

Fig 1 Flow Chart

VIII. CONCLUSION REFERENCES

In summary, combining MTCNN and FaceNet in facial [1]. Rajesh Kumar Tripathi, Anand Singh Jalal, Novel local
recognition offers a robust solution for identifying criminals feature extraction for age invariant face recognition,
and missing persons. By merging real-time face detection Expert Systems with Applications, Volume 175, 2021,
with precise feature extraction, this system effectively 114786, ISSN 0957-
overcomes challenges related to lighting, angles, and 4174https://doi.org/10.1016/j.eswa.2021.114786.
expressions, providing a more dependable tool for [2]. Shtwai Alsubai, Monia Hamdi, Sayed Abdel-Khalek,
identification. Experimental results indicate its effectiveness Abdullah Alqahtani, Adel Binbusayyis, Romany F.
in various situations, making it a valuable resource for law Mansour, Bald eagle search optimization with deep
enforcement. Moreover, ethical guidelines and privacy transfer learning enabled age-invariant face recognition
protections are integrated into the system, promoting model, Image and Vision Computing,Volume 126, 2022,
responsible usage. This research underscores the important 104545, ISSN 0262- 8856,
role of advanced technology in boosting public safety and https://doi.org/10.1016/j.imavis.2022.104545.
enabling timely, accurate investigations, all while upholding [3]. Jyothi S. Nayak, M. Indiramma. "An approach to
ethical principles and building public trust. enhance age invariant face recognition performance
based on gender classification." Journal of King Saud
University – Computer and Information Sciences, vol.
34, pp. 5183–5191, 2022.

IJISRT24DEC699 www.ijisrt.com 550

Volume 9, Issue 12, December – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165

[4]. Z. An, W. Deng, J. Hu, Y. Zhong, Y. Zhao. "APA:

Adaptive Pose Alignment for Pose-Invariant Face
Recognition." IEEE Access , vol. 7, pp. 14653-14668,
2019.
[5]. A. A M, A. G RAj, A. Devaraju C, S. B V. "Efficient
Tracking of Missing Person Using AI." International
Journal of Advanced Research in Computer and
Communication Engineering , vol. 12, issue 5, pp. 1090-
1094, 2023.
[6]. Roshin John , Basil Kuriakose. "Face Detection and
Tracking to Find the Missing Person," published in the
International Journal of Research in Engineering,
Science and Management , Volume 3, Issue 6, June
2020.
[7]. W. K. Aljohani, R. A. Alshehri, et al., "Suhail: A Deep
Learning-Based System for Identifying Missing
People." Computer and Information Science, Vol. 16,
No. 2; 2023.
[8]. S. Sambolek, M. Ivašić-Kos, "Person Detection in
Drone Imagery," Proceedings of the 2020 5th
International Conference on Smart and Sustainable
Technologies (SpliTech), 2020.
[9]. Roshin John , Basil Kuriakose. "Face Detection and
Tracking to Find the Missing Person," International
Journal of Research in Engineering, Science and
Management, Volume 3, Issue 6, June 2020. ISSN
(Online): 2581-5792.
[10]. V. Shelke, T. Bangera, et al., "Searchious: Locating
Missing People Using an Optimised Face Recognition
Algorithm." Proceedings of the 2021 5th International
Conference on Computing Methodologies (ICCMC),
2021

IJISRT24DEC699 www.ijisrt.com 551