LITERATURE SURVEY

BATCH : 2022 – 2024

ACADEMIC YEAR : 2023 – 2024

DEPARTMENT : MASTER OF COMPUTER APPLICATION

SUBJECT CODE/NAME : P20CA401/ PROJECT WORK

STUDENT DETAILS:

NAME OF THE PROJECT

REG NO NAME OF THE GUIDE
STUDENT BATCH NO

Ms. KALPANA,MCA.,
MOHAMMED Assistant Professor,
810422621028
ADHIL A Department of Master of
Computer Applications

TITLE OF THE PROJECT

WEAPON DETECTION SYSTEM USING FLASK WEB APP

1.TITLE: WEAPON DETECTION IN REAL-TIME CCTV VIDEOS USING DEEP
LEARNING

AUTHOR: MUHAMMAD TAHIR BHATTI, MUHAMMAD GUFRAN KHAN, MASOOD

ASLAM, MUHAMMAD JUNAID FIAZ1

PUBLISHED YEAR: 2021 PUBLISHER: IEEE Access

ALGORITHM: Deep learning, Object detection, Artificial Intelligence, Computer Vision

DESCRIPTION:

The innovative automatic weapon detection system introduced in this study marks a
significant advancement in real-time monitoring and control capabilities. With a primary
focus on enhancing security and maintaining law and order, this technology holds promise for
fostering a safer environment, particularly in regions afflicted by persistent violent activities.
By addressing the pressing need for effective detection mechanisms, this system stands to
bolster both the safety of communities and the stability of nations. One of the key
contributions of this work lies in its ability to mitigate false positives and negatives, thereby
ensuring the reliability and accuracy of weapon detection. Through meticulous construction
of a specialized training database tailored for real-time scenarios, coupled with rigorous
evaluation on cutting-edge deep learning models, this approach has yielded remarkable
results. Leveraging both sliding window/classification and region proposal/object detection
methodologies, the system demonstrates a superior performance in identifying weapons
within live CCTV streams.

ADVANTAGES:

 Improves security by detecting weapons in real-time.

 Boosts economy through increased investments and tourism.
 Minimizes false positives and negatives, enhancing efficiency.
 Achieves high precision and recall in weapon detection.
 Utilizes advanced technology like Yolov4 for effective detection.

DISADVANTAGES:

 Requires ongoing improvement for reducing errors.

 Limited to detecting weapons, may not address other threats.
 Demands significant computational resources for implementation.
  Raises ethical concerns regarding privacy and surveillance.

2.TITLE: HAWK-EYE: AN AI-POWERED THREAT DETECTOR FOR

INTELLIGENT SURVEILLANCE CAMERAS

AUTHOR: AHMED ABDELMOAMEN AHMED, MATHIAS ECHI

PUBLISHED YEAR: 2021 PUBLISHER: IEEE Access

ALGORITHM: Threat detector, mask Region Based Convolutional Neural Networks (R-
CNN), Internet of Things (IoT)

DESCRIPTION:

The paper introduces Hawk-Eye, an AI-powered threat detection system designed for real-
time video surveillance. This innovative prototype, deployed locally on the camera at the
edge, demonstrates high prediction accuracy and rapid response times. By leveraging the
Intel NCS 2 device alongside a Raspberry Pi 3, the system efficiently processes data without
relying on cloud computing. Utilizing state-of-the-art models like Mask R-CNN for instance
segmentation, Hawk-Eye identifies weapons in surveillance footage and facilitates swift
intervention by security personnel. With a user-friendly interface accessible both locally and
on the cloud, users can interact seamlessly with the system. Furthermore, the addition of a
motion detection module enhances real-time monitoring capabilities. Through rigorous
experimentation, the system showcases remarkable processing speeds, surpassing commercial
frames per second metrics. The team's commitment to open-source principles is evident, as
they share their code and dataset publicly, fostering collaboration and advancement in the
field.

ADVANTAGES:

 Rapid response times due to local deployment and efficient processing.

 Efficient data processing without reliance on cloud computing.
 Utilization of state-of-the-art models like Mask R-CNN for accurate identification.
 User-friendly interface accessible both locally and on the cloud.

DISADVANTAGES:

 Possible limitations in scalability and adaptability to larger datasets.

 Requires further experimentation to ensure robustness.
 Dependency on specific hardware configurations.
  Potential challenges in integrating with existing surveillance systems.

3.TITLE: GESTURE DETECTION AND RECOGNITION BASED ON OBJECT

DETECTION IN COMPLEX BACKGROUND

AUTHOR: RENXIANG CHEN AND XIA TIAN

PUBLISHED YEAR: 2023 PUBLISHER: MDPI Journals

ALGORITHM: Hand gesture recognition, Human–computer interaction (HCI), complex

background

DESCRIPTION:

In this study, they have developed an advanced gesture detection and recognition algorithm
that stands out for its speed, precision, and resilience in identifying gestures against intricate
backgrounds. The approach incorporates two key components: the efficient layer aggregation
network module and the CBAM attention mechanism module. By leveraging these modules,
we not only streamline the model's parameters but also bolster its capability to extract
features effectively, thus enhancing its ability to detect gestures amidst complex backgrounds.
they method outperforms existing techniques in gesture detection and recognition, a fact
validated through extensive experimentation. Looking ahead, in research aims to refine
recognition accuracy with fewer samples and improve the network's adaptability by
eliminating anchor boxes without compromising recognition accuracy. Furthermore, we are
keen to explore the application of our model in detection and recognition within various
contexts.

ADVANTAGES:

 Enhanced speed, precision, and resilience in gesture detection and recognition.

 Incorporation of efficient layer aggregation network
 Outperforms existing techniques in gesture detection
 Aims to refine recognition accuracy with fewer samples and improve network
adaptability without compromising accuracy.

DISADVANTAGES:

 May require substantial computational resources due to the advanced algorithm.

 Complexity in implementation and understanding for non-experts.
  Dependence on extensive experimentation for validation, which could be time-
consuming.

4.TITLE: A NOVEL APPROACH FOR OBJECT IDENTIFICATION WITH DEEP

REINFORCEMENT MACHINE LEARNING

AUTHOR: SAURABH TIWARI, DR. S VEENADHARI, DR. SANJEEV K GUPTA

PUBLISHED YEAR: 2021 PUBLISHER: IAEME Publication

ALGORITHM: CNN, Deep Reinforcement Learning, Object Identification

DESCRIPTION:

This paper introduces a novel approach to object identification in images using a

reinforcement learning model. The proposed model employs a reinforcement learning agent
to determine the optimal policy for placing a mask in the correct position to identify the
object within the image. Drawing upon the shallower concept of the faster RCNN model, the
primary objective is to achieve accurate and efficient object identification within minimal
execution time. By leveraging reinforcement learning, which enables the agent to calculate
rewards and adjust its policy using Q-learning, the model offers an efficient strategy to
adaptively learn the optimal identification policy. Unlike conventional methods, this approach
acknowledges that not every object follows a common search path, allowing the agent to
learn from its mistakes and experiences to refine the identification process. Implementation
of the model is carried out using Python Anaconda, facilitating efficient execution.
Experimental results demonstrate that the proposed method can accurately identify object
instances by analyzing 10 to 15 regions.

ADVANTAGES:

 Achieves accurate and efficient object identification within minimal execution time.
 Utilizes reinforcement learning to adaptively learn the optimal identification policy.
 Learns from mistakes and experiences, the identification process for varied objects
 Utilizes Python Anaconda for efficient execution.

DISADVANTAGES:

 Implementation may require a deeper understand of reinforcement learning concepts.

 The reinforcement learning agent may require substantial computational resources.
  Effectiveness may vary depending on the complexity.
 Performance may be sensitive to hyperparameters tuning.

5.TITLE: DEEP LEARNING ALGORITHMS-BASED OBJECT DETECTION AND

LOCALIZATION REVISITED

AUTHOR: SAFA RIYADH WAHEED, NORHAIDA MOHD SUAIB, MOHD SHAFRY

MOHD RAHIM, MYASAR MUNDHER ADNAN, A. A. SALIM

PUBLISHED YEAR: 2021 PUBLISHER: IOP Publishing

ALGORITHM: Deep Learning Convolutional Neural Network, Object Identification

DESCRIPTION:

This paper provides a thorough overview of the advancements in computer vision (CV)
fields, particularly focusing on object detection (OD) and localizations using deep learning
(DL) algorithms. DL-based ODs have demonstrated significant potential due to their robust
learning capabilities, effectively addressing challenges such as occlusions, scale
transformations, and background variations. As demands for OD technologies continue to
grow, there is a push towards developing comprehensive networks capable of localizing and
classifying objects simultaneously, facilitated by back-propagation training methods. Recent
models have prioritized high-performance prediction capacity, indicating ongoing progress in
the field. Despite advancements, the CV domain remains in its early stages, with numerous
avenues for further exploration and intensive study to gain deeper insights into OD
landscapes. In essence, this article serves as a detailed roadmap, delineating the progression
of neural networks and associated learning systems within the CV field, thereby facilitating
navigation and understanding of this rapidly evolving area of research.

ADVANTAGES:

 Robust learning capabilities of deep learning algorithms enhance object detection.

 object detection models show significant potential in addressing scale transformations
 Back-propagation methods facilitate the development of comprehensive network.

DISADVANTAGES:

 Despite advancements, computer vision remains in its early stages, suggesting

limitations in fully exploiting its potential.
  The complexity of deep learning algorithms may lead to high computational
requirements and resource intensiveness.

6.TITLE: A SURVEY ON POLICE PREVENTIVE ACTION TRACKING SYSTEM

USING AI

AUTHOR: DR.A. B. GAVALI, PHULE PRAVINKUMAR, NIMGIRE PRATHMESH,

PARAKHE GANESH, MANE SHUBHAM

PUBLISHED YEAR: 2023 PUBLISHER: IJSRCSEIT

ALGORITHM: Deep learning, Convolutional Neural Network (CNN), Moving object

detection

DESCRIPTION:

Artificial Intelligence (AI) holds immense promise in bolstering law enforcement efforts by
leveraging data analysis, pattern recognition, and insightful conclusions to thwart and identify
criminal activities. A pioneering initiative within this realm is the development of a Police
Preventive Action Tracking System powered by AI. This innovative system integrates AI
methodologies into police operations, facilitating preemptive measures against criminal
behavior. The advantages of such a system are multifaceted, including heightened situational
awareness, enhanced operational efficiency and effectiveness, reduced costs and risks, and
bolstered trust and satisfaction among communities. Nonetheless, the implementation of such
a system is not without its challenges and constraints. Issues such as the quality and
availability of data, technological infrastructure requirements, ethical and legal
considerations, and social and cultural barriers pose significant hurdles. Despite these
obstacles, the potential of AI in law enforcement remains a beacon of hope for safer, more
secure communities.

ADVANTAGES:

 Enhances situational awareness for police operations.

 Improves efficiency and effectiveness in crime prevention.
 Reduces costs and risks associated with law enforcement.
 Increases trust and satisfaction among the community.

DISADVANTAGES:
  Data quality and availability challenges may hinder effectiveness.
 Technology and infrastructure requirements may be complex and costly.
 Ethical and legal issues surrounding AI use in policing need careful consideration.
 Social and cultural barriers may affect acceptance and implementation.

7.TITLE: AUTOMATIC WEAPON DETECTION FOR SURVEILLANCE

APPLICATIONS

AUTHOR: SHREYAS GAWADE, DR. R. VIDHYA, DR. R. RADHIKA

PUBLISHED YEAR: 2022 PUBLISHER: SSRN

ALGORITHM: Deep learning, Convolutional Neural Network (CNN), Moving object

detection, Alarm system

DESCRIPTION:

A machine learning model has been developed to detect knives, small guns, long guns, and
other weapons by tracking objects in images and videos obtained from CCTVs. When a
weapon is detected, an alarm is triggered. The model exhibits an accuracy of approximately
85%. Further enhancements can be made by gathering more data of various weapons, both
labeled and unlabeled, and training the model with this data. Additionally, employing GPUs,
DSP, and other kits can enhance the model's performance. This model could potentially be
extended to detect other crimes such as robbery, thereby contributing to the reduction of
crime in society. Furthermore, its real-time application could prove instrumental in crime
prevention efforts.

ADVANTAGES:

 Detects weapons with 85% accuracy, potentially preventing violent incidents.

 Can be improved with more diverse data and advanced hardware.
 Could be extended to detect other crimes like robbery, aiding in reducing criminal.
 Instrumental in immediate response and crime prevention efforts.

DISADVANTAGES:

 Potential for misidentifying harmless objects as weapons.

 Continuous surveillance raises privacy issues and may infringe on individual rights.
 Focuses primarily on weapon detection, ignoring other aspects of crime prevention.
8.TITLE: WEAPON DETECTION USING ARTIFICIAL INTELLIGENCE AND
DEEP LEARNING FOR SECURITY APPLICATIONS

AUTHOR: MS. S. RAJALAKSHMI, MS. S. DHIVYASHREE, MS. K. BUSHYA,

MS.U.JANANI

PUBLISHED YEAR: 2024 PUBLISHER: IRJMETS

ALGORITHM: Moving object detection, Alarm system, YOLOv3 Training Model

DESCRIPTION:

Weapon detection using the YOLO (You Only Look Once) algorithm offers a potent solution
for bolstering security and public safety. The algorithm's real-time processing capabilities,
coupled with its accuracy and adaptability, render it a valuable asset in automatically
identifying weapons within images and video streams. YOLO's efficiency lies in its ability to
swiftly analyze images and video frames, making it applicable across a diverse array of
scenarios where prompt detection of weapons is imperative. Moreover, its precision in object
detection serves to minimize false alarms, ensuring dependable weapon identification. By
automating the detection process, YOLO streamlines security operations, diminishing the
necessity for manual inspection and thereby enhancing overall efficiency. The versatility of
YOLO extends its utility to numerous domains, encompassing surveillance, public safety, and
access control. Notably, the efficacy of weapon detection with YOLO hinges upon meticulous
dataset collection, annotation, and model training to achieve optimal detection accuracy.

ADVANTAGES:

 YOLO offers quick analysis of images and video frames, enabling rapid weapon
detection.
 YOLO's precision in object detection minimizes false alarms.
 Automation of the detection process streamlines security operations, reducing the
need for manual inspection.

DISADVANTAGES:

 Despite its accuracy, YOLO may still miss detecting certain weapons in complex.
  The use of YOLO for surveillance and public safety may raise privacy issues.
 Developing and deploying YOLO-based systems for weapon detection could entail
significant financial investments in hardware and software resources.

9.TITLE: WEAPON DETECTION USING YOLO V3 FOR SMART

SURVEILLANCE SYSTEM

AUTHOR: SANAM NAREJO, BISHWAJEET PANDEY, DORIS ESENARRO VARGAS,

CIRO RODRIGUEZ, M. RIZWAN ANJUM

PUBLISHED YEAR: 2021 PUBLISHER: HINDAWI

ALGORITHM: Moving object detection, Alarm system, YOLOv2 Training Model

DESCRIPTION:

Implemented in this study is the cutting-edge YOLO V3 object detection model, trained using
a curated dataset specifically tailored for weapon detection. The proposed model empowers
machines or robots with the ability to discern hazardous weaponry and promptly notify
human administrators upon detection of guns or firearms at the perimeter. Experimental
findings demonstrate the superior efficacy of the trained YOLO V3 over its predecessor,
YOLO V2, with the added benefit of reduced computational costs. There exists an urgent
imperative to enhance current surveillance capabilities by leveraging advanced resources to
monitor human operators' effectiveness. The advent of smart surveillance systems heralds a
transformative shift, facilitated by the increasingly accessible low-cost storage, robust video
infrastructure, and enhanced video processing technologies.

ADVANTAGES:

 YOLO V3 model enables accurate detection of weapons, boosting perimeter security.

 Immediate alerts to human administrators upon weapon detection enhance response.
 Outperforms YOLO V2 with better accuracy and reduced computational costs.
 Smart surveillance systems leverage low-cost storage and robust video infrastructure.

DISADVANTAGES:

 Increased surveillance may raise privacy issues for individuals

 Over-reliance on automated systems may lead to complacency or errors.
  Initial investment in advanced technology and infrastructure may be prohibitive for
some organizations.

10.TITLE: REAL-TIME CONCEALED OBJECT DETECTION FROM PASSIVE

MILLIMETER WAVE IMAGES BASED ON THE YOLOV3 ALGORITHM

AUTHOR: LEI PANG, HUI LIU, YANG CHEN, JUNGANG MIAO

PUBLISHED YEAR: 2020 PUBLISHER: MDPI

ALGORITHM: Concealed Object Detection, Passive Millimeter Wave

DESCRIPTION:

Diverging from conventional metal security gates and X-ray detectors, non-contact and non-
cooperative PMMW imagers have emerged as the preferred choice for security screenings in
expansive public venues. Simultaneously, the YOLO algorithm has gained prominence as a
highly efficient real-time detection method with substantial potential for advancement. This
study specifically concentrates on real-time detection of metal contraband from the human
body using PMMW images, employing a modest dataset and YOLOv3 algorithms. Training
was conducted on Yolov3-13 and Yolov3-53 target detection models with varying
convolutional layers, followed by an analysis of their respective strengths and weaknesses.
Comparative analysis was performed with SSD algorithms based on experimental results.
The findings demonstrate the efficacy of YOLOv3-based contraband detection methods for
PMMW images, meeting real-time detection demands amidst high passenger volumes.
Considering trade-offs between detection accuracy, speed, and computational resources, the
YOLOv3-53 model emerges as more advantageous and effective, even with limited sample
data. However, constrained by equipment capabilities, the availability of multi-source
PMMW image data remains limited, necessitating data augmentation and refinement.

ADVANTAGES:

 PMMW imagers offer convenient non-contact security screening.

 YOLOv3 detects metal contraband in real-time, suitable for busy venues.
 YOLOv3-53 balances accuracy and speed effectively.
 YOLOv3 requires minimal data for quick implementation.
DISADVANTAGES:

 Limited PMMW image data diversity hampers training.

 Equipment dependency limits PMMW technology scalability.
 Expanding datasets for diverse contraband types is challenging.