Anomaly Audio-Visual Detection for Smart Cities Security & Technology Efficiency Management |
|
| Mentor: | Javier Macías Guarasa |
| Email: | javier.maciasguarasa@uah.es |
| Phone: | (+34) 91885 6918 (+34) 607760186 |
| University: | University of Alcalá |
| Partner Host Institution: | |
| Keywords: | Anomaly Detection, Smart Cities, Audio-Visual Monitoring |
Anomaly Audio-Visual Detection for Smart Cities Security & Technology Efficiency Management
The GOT-Energy-TALENT proposal hereby presented by GEINTRA Group involves different tasks related to the monitoring and analysis of different indicators in the context of home and city surveillance in order to optimize the smart cities capabilities in terms of energy management and security. Works hereby focused thus involve two different important aims in the actual and future society: security and energy.
This proposal in the GOT-Energy-Talent call is oriented to support the transition to a reliable, sustainable and competitive smart city, within the Smart Cities & Communities area: sustainable development of urban areas is a challenge of key importance. It requires new, efficient, and user-friendly technologies and services, in which surveillance applications for people monitoring and analysis, is already a well-developed field in some other advanced societies such as the Asian one. The proposal hereby presented tackles the attainment of a commercial-scale solution with a high market potential that analyzes energy, transport and communications resources and use in which people audio-visual surveillance is the main key.
The overall objective of the proposal focuses on understanding what is happening in a scene under surveillance, because of its interest in security and efficiency of smart-cities technology, including both activities and behavior of groups and individuals. The main target is automatically processing the different scenarios of interest and identifying the typical and atypical (usual or anomalous) activities that take place there, in order to provide further assesment to perform security and energy efficiency analysis.
Thus, the main subtopics of the proposal aim to identify relevant behaviour patterns of human activity in a given sensed area (either indoor or outdoor). These patterns will be related to normal and abnormal (anomalous) behaviours, so that automatic anomaly detection can be addressed. In this detection, acoustic and visual clues will be included, to provide a better understanding of the human activity, in which visualy detected activities and human vocal information play a fundamental role.
From a technological and algorithmic point of view, this will be achieved by combining distributed sensors and machine learning strategies. Distributed sensorization will focus on audio, video and depth information streams, and the information provided by them will be exploited by combining traditional pattern recognition approaches, and also current deep learning strategies. The proposal will first focus on the estimation of the statistical and structural properties of normal behaviours in the sensed scenes, and then developing specific methods to address anomaly detection and identificacion.
| Departament: | Electronics |
| Research Group: | GEINTRA |
| More Information: | http://www.geintra-uah.org/ http://www.uah.es/es/estudios/profesor/Javier-Macias-Guarasa/ |
| Relevants projects on the area: | HEIMDAL: "Multisensory semantic detection of anomalous situations in environments without restrictions" Ref: TIN2016-75982-C2-1-R Ministry of Economy and Competitiveness 30 Dec 2016 - 29 Dec 2020 |
| Relevants publications on the area: | 1.- J. Vera-Diaz, D. Pizarro, y J. Macias-Guarasa, «Towards End-to-End Acoustic Localization Using Deep Learning: From Audio Signals to Source Position Coordinates», Sensors, vol. 18, n.º. 10, p. 1-22, 2018 2.- C. A. Luna, C. Losada, D. Fuentes-Jimenez, A. Fernandez-Rincon, M. Mazo, y J. Macias-Guarasa, «Robust People Detection Using Depth Information from an Overhead Time-of-Flight Camera», Expert Systems with Applications, vol. 71, p. 240-256, 2017. |
