التفاصيل البيبلوغرافية
| العنوان: |
Weapon detection in real-time CCTV videos using deep learning. |
| المؤلفون: |
Mohiuddin, Mohammed Ahmed, Laxmikanth, M., Vardhani, G. Harsha, Hari, B. Damodhara, Rani, K. Swapna, Mouli, K. Chandra, Dhyani, Saurabh |
| المصدر: |
AIP Conference Proceedings; 2024, Vol. 3101 Issue 1, p1-7, 7p |
| مصطلحات موضوعية: |
MACHINE learning, DEEP learning, INTERNATIONAL tourism, CLOSED-circuit television, WEAPONS, FOREIGN investments |
| مستخلص: |
Modern world places a high value on safety & security. How safe & secure an environment is determines a nation's ability towards attract tourism & foreign investment. However, despite fact certain Closed-Circuit Television (CCTV) cameras are utilized for surveillance & towards keep an eye on events such as robberies, they still require human intervention & oversight. A system certain can quickly identify these crimes is what we need. Indeed, even among state about art profound learning calculations, speedy handling power, & modern CCTV cameras, ongoing weapon identification stays a critical obstruction. Variable viewing angles, occlusions caused through firearm's owner & those nearby, & other factors further complicate exercise. Using cutting-edge open-source deep learning algorithms, aforementioned effort aims towards create a safe environment through using CCTV footage as a source towards identify dangerous weapons. We developed relevant confusion items inclusion ideas & built a binary classification using pistol class as reference class towards eliminate false positives & false negatives. We created our own because there was no standard dataset for real-time scenario. We did aforementioned through shooting weapons among our own cameras, manually collecting images from internet, extracting data from YouTube CCTV recordings, & making use about GitHub repositories. two approaches used are sliding window/classification & region proposal/object detection. calculations YOLOV5, YOLOV6, YOLOV&, & Quicker RCNN are a not many certain are utilized. While performing object identification, accuracy & review matter more than exactness, consequently full arrangement about techniques was assessed as far as these. Yolov5 stands out most among other algorithms among a F1-score about 91 percent & a mean average precision certain is 91.73 percent higher than before. [ABSTRACT FROM AUTHOR] |
|
Copyright of AIP Conference Proceedings is the property of American Institute of Physics and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.) |
| قاعدة البيانات: |
Complementary Index |
Find this article in full text from Scitation.