Scholarlink

In the current era, providing security is being a top priority in almost every domain, due to a rise in crime rates in crowded events or vulnerable lonely areas. Many of the attacks can be avoided if the attack can be predicted beforehand. The major number of attacks are been from the attackers holding the weapons with hand, hence finding the hidden weapons with the persons in the surveillance videos or images can be more helpful for saving from unforeseen attacks. With the increasing demand for intelligent security surveillance cameras, different studies on a detection system in terms of pre-processing, manipulation, and interpretation of identified video frames and images have been conducted. In this work, weapon detection of different handguns in different complex scenarios through videos and images were implemented. There is a notable need for the neutralization of terrorist attacks, and the fastest and earliest detection of such threats is a major area of concern to ensure human safety. Everywhere there is a need to deploy these types of models for smart surveillance cameras and security systems. The detection procedure can include analysis of both static and dynamic pictures and images to constantly monitor the area for security. In this paper, we are using a deep learning CNN algorithm to detect weapons like knives and handguns from images or videos and keeping an eye of an object and its related issues, such as extraction of object features, analysis of the shape of object, and analyzing the position of the object. To carry out proposed work, we have used some weapons and sharp objects’ datasets which are already labelled and some are manually labelled that contain handguns, pistols, and knives’ images. The proposed implementation uses two types of datasets. The model was compared with some existed objects’ detection data models against some chosen parameters. The results are plotted and tabulated. With the proposed model, good accuracy was achieved in detection of the weapons in both videos and images. The speed of detection is also significantly high in this model.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

An Approach for Recognizing Object Using Convolutional Neural Networks for Detection of Weapons

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • N. Sowjanya Kumari,
  • Sreelahari Vallamsetla,
  • Ravi Ravva,
  • Sekharmahanti S. Nandini

摘要

In the current era, providing security is being a top priority in almost every domain, due to a rise in crime rates in crowded events or vulnerable lonely areas. Many of the attacks can be avoided if the attack can be predicted beforehand. The major number of attacks are been from the attackers holding the weapons with hand, hence finding the hidden weapons with the persons in the surveillance videos or images can be more helpful for saving from unforeseen attacks. With the increasing demand for intelligent security surveillance cameras, different studies on a detection system in terms of pre-processing, manipulation, and interpretation of identified video frames and images have been conducted. In this work, weapon detection of different handguns in different complex scenarios through videos and images were implemented. There is a notable need for the neutralization of terrorist attacks, and the fastest and earliest detection of such threats is a major area of concern to ensure human safety. Everywhere there is a need to deploy these types of models for smart surveillance cameras and security systems. The detection procedure can include analysis of both static and dynamic pictures and images to constantly monitor the area for security. In this paper, we are using a deep learning CNN algorithm to detect weapons like knives and handguns from images or videos and keeping an eye of an object and its related issues, such as extraction of object features, analysis of the shape of object, and analyzing the position of the object. To carry out proposed work, we have used some weapons and sharp objects’ datasets which are already labelled and some are manually labelled that contain handguns, pistols, and knives’ images. The proposed implementation uses two types of datasets. The model was compared with some existed objects’ detection data models against some chosen parameters. The results are plotted and tabulated. With the proposed model, good accuracy was achieved in detection of the weapons in both videos and images. The speed of detection is also significantly high in this model.