TY - GEN
T1 - VidSearch
T2 - 22nd IEEE International Conference on Machine Learning and Applications, ICMLA 2023
AU - Tahir, Mehwish
AU - Qiao, Yuansong
AU - Kanwal, Nadia
AU - Lee, Brian
AU - Asghar, Mamoona Naveed
N1 - Publisher Copyright:
© 2023 IEEE.
PY - 2023
Y1 - 2023
N2 - The surge in surveillance camera deployment in the era of Big Data and the Internet of Things (IoT) has emphasized the paramount importance of safeguarding the privacy of individuals, objects, and locations they record. Therefore, this paper proposes VidSearch - a secure system designed for storing, searching, and retrieving videos captured by CCTV cameras. VidSearch system enhances visual data protection through encryption, query-by-text video searching within encrypted data, and anonymized video retrieval using pixelization. During storage, encrypted videos and their metadata are stored separately to facilitate text-based search and video retrieval from encrypted videos. Fernet encryption is applied to protect videos, and two anonymization algorithms i.e., a Mixture of Gaussians 2 (MOG2) and K-Nearest Neighbors (KNN) are used for detecting the foreground (moving objects) and background of the videos at the retrieval stage. Video retrieval results demonstrate that KNN excels in accuracy for visual content detection, while MOG2 is more efficient in terms of processing time. VidSearch system is extensively tested on a general-purpose Intel system and an IoT NVIDIA Jetson. Results confirm the system's ability to operate in a Big Data and IoT ecosystem across multiple devices and platforms.
AB - The surge in surveillance camera deployment in the era of Big Data and the Internet of Things (IoT) has emphasized the paramount importance of safeguarding the privacy of individuals, objects, and locations they record. Therefore, this paper proposes VidSearch - a secure system designed for storing, searching, and retrieving videos captured by CCTV cameras. VidSearch system enhances visual data protection through encryption, query-by-text video searching within encrypted data, and anonymized video retrieval using pixelization. During storage, encrypted videos and their metadata are stored separately to facilitate text-based search and video retrieval from encrypted videos. Fernet encryption is applied to protect videos, and two anonymization algorithms i.e., a Mixture of Gaussians 2 (MOG2) and K-Nearest Neighbors (KNN) are used for detecting the foreground (moving objects) and background of the videos at the retrieval stage. Video retrieval results demonstrate that KNN excels in accuracy for visual content detection, while MOG2 is more efficient in terms of processing time. VidSearch system is extensively tested on a general-purpose Intel system and an IoT NVIDIA Jetson. Results confirm the system's ability to operate in a Big Data and IoT ecosystem across multiple devices and platforms.
KW - Privacy Preservation
KW - Query-by-Text Search
KW - Surveillance Systems
KW - Video Retrieval
KW - Video Storage
UR - http://www.scopus.com/inward/record.url?scp=85190160446&partnerID=8YFLogxK
U2 - 10.1109/ICMLA58977.2023.00329
DO - 10.1109/ICMLA58977.2023.00329
M3 - Conference contribution
AN - SCOPUS:85190160446
T3 - Proceedings - 22nd IEEE International Conference on Machine Learning and Applications, ICMLA 2023
SP - 2182
EP - 2187
BT - Proceedings - 22nd IEEE International Conference on Machine Learning and Applications, ICMLA 2023
A2 - Arif Wani, M.
A2 - Boicu, Mihai
A2 - Sayed-Mouchaweh, Moamar
A2 - Abreu, Pedro Henriques
A2 - Gama, Joao
PB - Institute of Electrical and Electronics Engineers Inc.
Y2 - 15 December 2023 through 17 December 2023
ER -