#	Pagina
attuale pagina	/open-h2020/projects/203260/results.html

Report

Teaser, summary, work performed and final results

Periodic Reporting for period 1 - SCENT (Smart Toolbox for Engaging Citizens into a People-Centric Observation Web)

Teaser

Summary

SCENT provides an integrated toolbox of smart collaborative and innovative technologies that augment costly in situ infrastructure, enabling citizens to become the â€˜eyesâ€™ of the policy makers. SCENT enables citizens to monitor Land Cover and Land Use (LC/LU) changes, in their everyday activities, and related environmental phenomena, such as floods, by crowdsourcing relevant information. Two large scale demonstrations are planned in order to showcase the usefulness of the toolbox in real but diverse examples where flood monitoring is a great importance. The first region is the wetland of Danube Delta in Romania, where flooding is a necessary and welcome part of maintaining the ecosystem, and the second is the Kifisos River basin around the urban landscape of Athens in Greece, where flooding is a major problem for the local communities.

Work performed

Despite being in the middle of the project workplan, the project has quite significant results so far:
Through a constant interaction with stakeholders in the field of environmental monitoring, user needs and requirements for an efficient citizen observatory on land resource monitoring were elicited, since the early steps of the project. Following this, the SCENT toolbox system requirements and architecture were defined, aiming to respond to these needs and to allow the deployment of the toolbox to the pilot areas.
The development of SCENT crowdsourcing and gaming applications was initiated and is in a mature stage of development. The SCENT Explore app enables users to visit areas of interest, take photos of LC/LU and other important environmental parameters and tag them. At the moment, several functionalities, such as the development of a user-friendly interface, the gamification dynamics, and the annotation mechanism, have been implemented, with the focus in the coming months being set mainly to the full integration with the backend of the platform and the completion of added value functions (i.e. signal of events of interest, visualization of such events on maps, etc). In addition, SCENT Collaborate application, and CAPTCHA plugin are developed aiming to establish a validation process through which user annotations and automated annotations from the SCENT toolbox automated image classification tool, i.e. SCENT Intelligent Engine (SIE), can be confirmed. This validation will be used to improve the existing training dataset for SIE and calculate the user trust level. Both applications will be fully operational by the end of the 2nd year of the project.
The design of SCENT crowdsourcing platform was completed, identifying also the functionalities supported by its core components. One of them, the crowdsourcing backend, has already been established as key data exchange for all citizen-generated data, connecting the SCENT crowdsourcing frontends (SCENT Explore app, captcha, SCENT Collaborate) with other core parts of the toolbox (i.e. SIE and with the harmonisation platform). Moreover, a web-based application- the SCENT Authoring tool has been developed, aiming to support relevant authorities (i.e. policy makers, flood managers, etc) in creating crowdsourcing campaigns targeted at specific areas where images and sensor measurements are requested (e.g. measuring parameters of the river, taking images of urban obstacles etc). Thus, volunteers can visit these points of interest and collect the necessary information.
The SCENT in-situ, portable and ready-to-fly digital sensors have been defined obtained and installed, aiming to collect environmental measurements that cover air, soil and water parameters. At the moment, a sensor measurement collection app is being developed that will allow the collection and visualization of sensor data and the integration of the latter with the SCENT Crowdsourcing platform. A Data Quality Control mechanism is also being implemented, coping with the correctness/accuracy of the sensor measurements and user reliability.
In order to augment costly in-situ sensors and collect river measurements with a consistent and accurate way from volunteers, two tools have been designed that enable the extraction of the water level and water velocity from images and videos respectively.
The architecture of the SCENT Intelligence Engine has been designed and a large set of relevant images has been collected from open image platforms, manually annotated and used in training and testing of its core image analysis tool. Innovative machine learning techniques are used to extract all the information relevant to the Scent taxonomy from the images.
A set of hydrodynamic and hydrological models of Danube Delta and Kifisos pilot cases have been developed using existing in-situ data. In the upcoming months, methods will be realized to enable the integration of crowdsourced data into the models.
The information c

Final results

Until the completion of the project, SCENT envisions significant progress beyond the state of the art for the various components of the toolbox, whilst worth highlighting some of them as follows:
Development of a crowd-sourcing platform that will cover the entire spectrum of collection, interpretation and validation of user generated data;
Implementation of a machine learning based image classification will combine categorical classification of scenes (field, forest, etc) with land-cover object type recognizers (e.g. types of grass, etc) to produce a superior scene description engine able to provide environmental data;
Maximization of user engagement through the use of a gamification mechanism and a reward system that will motivate users into collecting and annotating new images from the pilot areas;
Improved flood risk and flood pattern mapping through the provision of a methodology by which crowd sourced data can be combined with available ground observation and remote sensing that improve physically based models used for flood mapping.

The impact of the projectâ€™s activities and outcomes until the end of the project can be summarised as follows:
Lowered cost and extension of the in-situ component of the GEOSS and Copernicus initiatives;
Better decision-making through the empowerment and active role of citizens and citizenâ€™s associations in environmental monitoring;
Enhanced implementation of governance and global policy objectives;
Increased deployment and market uptake of innovative in-situ monitoring techniques;
Increased European role in the business of in-situ monitoring of the environment;
Fostering innovation in SMEs in order to combat the increasing rates of European unemployment