Ancilia: Scalable Intelligent Video Surveillance for the Artificial Intelligence of Things

Pazho, Armin Danesh; Neff, Christopher; Noghre, Ghazal Alinezhad; Ardabili, Babak Rahimi; Yao, Shanle; Baharani, Mohammadreza; Tabkhi, Hamed

doi:10.48550/arxiv.2301.03561

Cited by 3 publications

(5 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Simultaneously, the video frame was sent in the suggested queue to request for detecting and tracking objects. Several efforts in [14][15][16][17] on anomalous activity monitoring were based mostly on low-level features. However, all works above considered a policy of always loading a heavy DL model that tracked the movement of an object in the DL-based video surveillance system, which results in unnecessary GPU memory consumption when there is no object appearance.…”

Section: Related Workmentioning

confidence: 99%

Cognitive Video Surveillance Management in Hierarchical Edge Computing System with Long Short-Term Memory Model

Ugli

Kim²,

Mohammed

et al. 2023

Sensors

View full text Add to dashboard Cite

Nowadays, deep learning (DL)-based video surveillance services are widely used in smart cities because of their ability to accurately identify and track objects, such as vehicles and pedestrians, in real time. This allows a more efficient traffic management and improved public safety. However, DL-based video surveillance services that require object movement and motion tracking (e.g., for detecting abnormal object behaviors) can consume a substantial amount of computing and memory capacity, such as (i) GPU computing resources for model inference and (ii) GPU memory resources for model loading. This paper presents a novel cognitive video surveillance management with long short-term memory (LSTM) model, denoted as the CogVSM framework. We consider DL-based video surveillance services in a hierarchical edge computing system. The proposed CogVSM forecasts object appearance patterns and smooths out the forecast results needed for an adaptive model release. Here, we aim to reduce standby GPU memory by model release while avoiding unnecessary model reloads for a sudden object appearance. CogVSM hinges on an LSTM-based deep learning architecture explicitly designed for future object appearance pattern prediction by training previous time-series patterns to achieve these objectives. By referring to the result of the LSTM-based prediction, the proposed framework controls the threshold time value in a dynamic manner by using an exponential weighted moving average (EWMA) technique. Comparative evaluations on both simulated and real-world measurement data on the commercial edge devices prove that the LSTM-based model in the CogVSM can achieve a high predictive accuracy, i.e., a root-mean-square error metric of 0.795. In addition, the suggested framework utilizes up to 32.1% less GPU memory than the baseline and 8.9% less than previous work.

show abstract

Section: Related Workmentioning

confidence: 99%

Cognitive Video Surveillance Management in Hierarchical Edge Computing System with Long Short-Term Memory Model

Ugli

Kim²,

Mohammed

et al. 2023

Sensors

View full text Add to dashboard Cite

show abstract

“…Fig 2 shows the AI algorithm pipeline on the local node and global node used for communication with the cloud node in the edge. [10] For each camera one AI algorithm pipeline is assigned for processing. The local node deep learning pipeline commences by extracting images from the surveillance cameras and downsampling their resolution to 720p to ensure consistent processing rates across all cameras in the loop.…”

Section: System Featuresmentioning

confidence: 99%

“…In order to reduce the load on the cloud node side, all the data sent to the cloud node is run through statistical analysis to generate better statistical anomaly data and action anomaly data. More details can be found in [10] A. Cloud Node…”

Section: System Featuresmentioning

confidence: 99%

“…It aimed to address the challenges above in an environment with a comparatively lower likelihood of anomalous behavior than in high crime rate environments. In this study, we utilized a designed scalable SVS system [10] to conduct the Artificial Intelligence (AI) vision analysis and detect anomalous behaviors. We also employed a smartphone application [11] to push notifications to the end user in case of detecting anomalies.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Real-World Community-in-the-Loop Smart Video Surveillance -- A Case Study at a Community College

Yao¹,

Ardabili²,

Pazho³

et al. 2023

Preprint

View full text Add to dashboard Cite

Smart Video surveillance systems have become important recently for ensuring public safety and security, especially in smart cities. However, applying real-time artificial intelligence technologies combined with low-latency notification and alarming has made deploying these systems quite challenging.This paper presents a case study for designing and deploying smart video surveillance systems based on a real-world testbed at a community college. We primarily focus on a smart camerabased system that can identify suspicious/abnormal activities and alert the stakeholders and residents immediately. The paper highlights and addresses different algorithmic and system design challenges to guarantee real-time high-accuracy video analytics processing in the testbed. It also presents an example of cloud system infrastructure and a mobile application for realtime notification to keep students, faculty/staff, and responsible security personnel in the loop. At the same time, it covers the design decision to maintain communities' privacy and ethical requirements as well as hardware configuration and setups.We evaluate the system's performance using throughput and end-to-end latency. The experiment results show that, on average, our system's end-to-end latency to notify the end users in case of detecting suspicious objects is 5.3, 5.78, and 11.11 seconds when running 1, 4, and 8 cameras, respectively. On the other hand, in case of detecting anomalous behaviors, the system could notify the end users with 7.3, 7.63, and 20.78 seconds average latency. These results demonstrate that the system effectively detects and notifies abnormal behaviors and suspicious objects to the end users within a reasonable period. The system can run eight cameras simultaneously at a 32.41 Frame Per Second (FPS) rate.

show abstract

“…Recently smart cities have been an emerging topic, drawing the attention of scholars in various fields [7], [8]. Among the essentials of a "Smart City" is transit reliability, which is imperative for commuters seeking to eliminate lengthy commutes and public transit wait times.…”

Section: Introductionmentioning

confidence: 99%

A Novel Neural Network Approach for Predicting the Arrival Time of Buses for Smart On-Demand Public Transit

Rashvand¹,

Hosseini²,

Azarbayjani³

et al. 2023

Preprint

View full text Add to dashboard Cite

Among the major public transportation systems in cities, bus transit has its problems, including more accuracy and reliability when estimating the bus arrival time for riders. This can lead to delays and decreased ridership, especially in cities where public transportation is heavily relied upon. A common issue is that the arrival times of buses do not match the schedules, resulting in latency for fixed schedules. According to the study in this paper on New York City bus data, there is an average delay of around eight minutes or 491 seconds mismatch between the bus arrivals and the actual scheduled time. This research paper presents a novel AI-based data-driven approach for estimating the arrival times of buses at each transit point (station). Our approach is based on a fully connected neural network and can predict the arrival time collectively across all bus lines in large metropolitan areas. Our neural-net data-driven approach provides a new way to estimate the arrival time of the buses, which can lead to a more efficient and smarter way to bring the bus transit to the general public. Our evaluation of the network bus system with more than 200 bus lines, and 2 million data points, demonstrates less than 40 seconds of estimated error for arrival times. The inference time per each validation set data point is less than 0.006 ms.

show abstract

Ancilia: Scalable Intelligent Video Surveillance for the Artificial Intelligence of Things

Cited by 3 publications

References 0 publications

Cognitive Video Surveillance Management in Hierarchical Edge Computing System with Long Short-Term Memory Model

Cognitive Video Surveillance Management in Hierarchical Edge Computing System with Long Short-Term Memory Model

Real-World Community-in-the-Loop Smart Video Surveillance -- A Case Study at a Community College

A Novel Neural Network Approach for Predicting the Arrival Time of Buses for Smart On-Demand Public Transit

Contact Info

Product

Resources

About