User Tools

Site Tools


projects:year10:10a.002.tau_wp3

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
Next revision
Previous revision
projects:year10:10a.002.tau_wp3 [2021/05/11 08:45]
sally.johnson [Table]
projects:year10:10a.002.tau_wp3 [2022/10/03 14:57] (current)
sally.johnson [Project - Benefits to IAB]
Line 3: Line 3:
 <WRAP leftalign box > <WRAP leftalign box >
  
-  * Natural catastrophes and outbreaks, such as COVID-19, impose restrictions on citizens movement and daily life. Such restrictions are maintained by governmental bodies and agencies using CCTV cameras, monitored by operators, and by on site personnelHoweverthese conventional monitoring techniques are very labor intensive and suffer from subjective interpretations and human error due to fatigue. +The rapid outbreak of the Coronavirus Disease 2019 (COVID-19) has imposed restrictions on people’s movement and daily life [1]Reducing the spread of the virus mandates constraining social interactions, traveling, and access to public areas and events [1]. These limitations arise to mainly advocate social distancing; the practice of increasing physical space among people to minimize virus transmission [2]. Monitoring and maintaining social distancing are carried out by governmental bodies and agencies using mass surveillance systems and closed-circuit television (CCTVcameras [3]Nonethelessthis task is cumbersome and suffers from subjective interpretations and human error due to fatigue; hence, computer vision and machine learning tools are convenient for automation [4]In addition, they enable crowd behavior to be monitored and anomalies such as congested regions, curfew infractions, and illegal gatherings to be recognized. The widespread of mass surveillance and its integration with Machine Learning is hindered by ethical concerns, including possible breach of privacy and potential abuse [3]. Therefore, privacy-preserving surveillance and Machine Learning solutions are paramount to their ethical adoption and application [5]. The design of vision-based social distance estimation and crowd monitoring system deals with the following challenges [4]: (1) geometry understanding, in terms of ground plane identification and homography estimation; (2) multiple people detection and localization; and (3) statistical/temporal characterization for social distance infractions, e.g., short-term violations are irrelevant. Currently, Machine Learning-based solutions identify social distance infringements using off-the-shelf person detection and tracking models [4]. In general, the models’ performance is conjoined with privacy; they yield high performance by carrying and processing person-specific information to develop robustness against occlusions and missing data [4]. In addition, they localize human subjects via bounding boxes that can be over-sized or incomplete which results in significant distance estimation errors [6]. Therefore, we developed a privacy-preserving adaptive social distance estimation and crowd monitoring system that can be implemented on top of any existing CCTV infrastructure. Specifically, we designed a novel person localization strategy through pose estimation, built a privacy-preserving adaptive smoothing and tracking model to mitigate occlusions and noisy/missing measurementscomputed inter-personal distances in the real-world coordinates, detected social distance infractionsand identified overcrowded regions in a scene. Performance evaluation was carried out by testing the system’s ability in person detection, localization, density estimation, anomaly recognition, and high-risk areas identification. We compared the proposed system to the latest techniques and examined the performance gain delivered by the localization and smoothing/tracking algorithms. Experimental results indicated a considerable improvementacross different metricswhen utilizing the developed system. In additionthey showed its potential and functionality for applications other than social distancingIn brief, the main contributions of this project are as follows: (1) developing a robust person localization strategy using pose estimation techniques; (2) forming an adaptive smoothing and tracking paradigm to mitigate the problem of occlusions and missing data without compromising privacy; (3) designing a real-time privacy-preserving social distance estimation and crowd monitoring solution with potential to cover other application areas and tasks.
-  * This project aims to provide a real-time early anomaly recognition system based on advanced Computer Vision and Deep Learning algorithms that can be implemented on top of a wide CCTV infrastructure and monitoring grid. +
-  * The proposed system aims to detect and identify curfew infractions, social distancing violationsillegal gatherings, and general threats such as firesmokeunattended objects in public places, and abnormal behaviors. +
-  * The application domains of this project include both surveillance and empathic buildings development.+
  
  
 </WRAP> </WRAP>
 ===== Project - Team ===== ===== Project - Team =====
-^ Team Member       ^ Role        ^ Email                    ^ Phone Number     ^ Academic Site/IAB                            +^ Team Member                                          ^ Role            ^ Email                      ^ Phone Number     ^ Academic Site/IAB   
-| Moncef Gabbouj    | PI          | moncef.gabbouj@tuni.fi   | 358 40 073 6613  | Tampere University                           +| [[about:personnel:moncef_gabbouj|Moncef Gabbouj]]    | PI              | moncef.gabbouj@tuni.fi     | 358 40 073 6613  | Tampere University  
-| Serkan Kiranyaz   | Co-PI       | serkan.kiranyaz@tuni.fi  | 97 43 063 5600   | Tampere University                           +| [[about:personnel:serkan_kiranyaz|Serkan Kiranyaz]]  | Co-PI           | serkan.kiranyaz@tuni.fi    | 97 43 063 5600   | Tampere University  
-| Mohammad Al-Sa' | Researcher  | mohammad.al-sad@tuni.fi  | 358 41 799 3159  | Tampere University                           +| [[about:personnel:mohammad_alsad|Mohammad Al-Sa'd]]  | Researcher      | mohammad.al-sad@tuni.fi    | 358 41 799 3159  | Tampere University  
-                  |                                                       **Funded by: Haltian & Tampere University**  |+Matti Vakkuri                                        Project Mentor  | matti.vakkuri@haltian.com  | N/A              | Haltian             | 
 +Tomi Teikko                                          | Project Mentor  | N/A                        | N/A              | Haltian             | 
 +| Gunnar Hansen                                        | Project Mentor  | N/A                        | N/A              | Haltian             |
 ===== Project - Novelty of Approach ===== ===== Project - Novelty of Approach =====
 <WRAP leftalign box > <WRAP leftalign box >
  
-  * Detecting and identifying threats and abnormal behaviors in video feeds have been a hot topic ever since computer vision algorithms recently became popular thanks to the advances made in deep learning; however, no convincing real-time solution has been provided up to date. +  * Detecting and identifying threats and abnormal behaviors in video feeds have been a hot topic ever since computer vision algorithms became popular thanks to the advances made in deep learning; however, no convincing real-time solution has been provided up to date. In this project, we designed a real-time crowd monitoring and anomaly detection systems that can be deployed on CPU and GPU machines
-  * Current anomaly detection solutions excel given specific anomalies and conditions. Nonetheless, we aim to leverage various optimized techniques to yield a comprehensive early anomaly recognition system.+  * Current anomaly detection solutions excel given specific anomalies and conditions. Therefore, we leveraged various optimized techniques to yield a comprehensive social distance estimation and crowd monitoring system. 
 +  * Current anomaly detection systems yield high performance by carrying and processing person-specific information to develop robustness against occlusions and missing data. In addition, they localize human subjects via bounding boxes that can be over-sized or incomplete which results in significant distance estimation errors. We solved these shortcomings by using a pose estimation technique to detect people because it is independent of the subject’s height, width, and orientation and carries no person-specific information; hence, it preserves privacy. 
  
 </WRAP> </WRAP>
 ===== Project - Deliverables ====== ===== Project - Deliverables ======
  
-^    ^ Deliverables                                                                  +^    ^ Deliverables                                                     
-| 1  | Video feeds from different cameras optimized for efficient anomaly detection  +| 1  | Identifying and collecting suitable datasets                     
-| 2  | Software based on existing methods written in MATLAB/Python/C++               +| 2  | Proof of concept of the early warning system                     
-| 3  | Software utilizing GPUs and parallel computation for enhanced performance     | +| 3  | System integration, verification and publication of the results  |
-| 4  | Multi-view early anomaly recognition system framework                         | +
-| 5  | Quantitative investigation identifying strengths and potential weaknesses     |+
 ===== Project - Benefits to IAB ===== ===== Project - Benefits to IAB =====
 <WRAP leftalign box > <WRAP leftalign box >
  
-The recent COVID-19 outbreak imposes immediate needs for such a comprehensive surveillance and tracking system for indoors and outdoors. The application domains include both surveillance and empathic buildings development. This is beneficial for any CVDI company who wishes to be among the pioneers of this next generation monitoring systems and empathic building.+The COVID-19 outbreak imposed immediate needs for such a comprehensive surveillance and tracking system for indoors and outdoors. The application domains include both surveillance and empathic buildings development. Moreover, apart from the project direct application to anomaly detection, social distance estimation, and crowd monitoring, the designed system occupancy/crowd density map functionality extends its application domain beyond the COVID-19 pandemic to cover other areas. For instance, it can help re-configure or re-design common physical layouts and relocate facilities in businesses to optimally reduce congestion. Additionally, it can facilitate the analysis of customer’s browsing habits in shops and quantifying the effectiveness of marketing kiosks.
  
 </WRAP> </WRAP>
Line 43: Line 42:
 {{ :meetings:spring2021:10a.002.tau_wp3_executive_summary_04.09.2021.docx |}}\\ {{ :meetings:spring2021:10a.002.tau_wp3_executive_summary_04.09.2021.docx |}}\\
 {{ :meetings:spring2021:10a.002.tau_wp3_pitch_slides.pptx |}}\\ {{ :meetings:spring2021:10a.002.tau_wp3_pitch_slides.pptx |}}\\
 +{{ :projects:year10:10a.002.tau_wp3_update_presentation_fall_meeting_2021_updated.pdf |}}\\
 +{{ :projects:year10:10a.002.tau_wp3_update_presentation_fall_meeting_2021_updated.pptx |}}\\
 +{{ :projects:year10:10a.002.tau.wp3_project_update_spring_2022.pptx |}}\\
 +{{ :projects:year10:10a.002.tau_wp3_final_report.pdf |}}\\
  
 ~~DISCUSSION~~ ~~DISCUSSION~~
projects/year10/10a.002.tau_wp3.1620740706.txt.gz · Last modified: 2021/05/11 08:45 by sally.johnson