The SARUAV system for detecting lost persons in aerial imagery acquired by unmanned aerial vehicles

Name of the provider (company name or main contact name), or FIRE IN ID ? SARUAV sp. z o.o.

Scope, rationale, context: general description. Precise here if this technology is currently use (eg. company name or contact info) The SARUAV system is software that provides spatial models and processing algorithms for detecting lost persons in the wilderness. It is a fully research-based solution, dedicated for search and rescue services. It processes aerial imagery acquired by unmanned aerial vehicles and automatically detects persons in the wilderness in order to reduce search time and therefore to increase the probability of survival. The software has been developed by the spin-off company SARUAV sp. z o.o. (www.saruav.pl) which works under the patronage of the University of Wrocław (Poland), where a prototype of the solution has been elaborated. The system has been subject to numerous tests, including field experiments. Since June 2020 the SARUAV system has been available on the market. The licences for its operational use have been purchased by three search and rescue organisations: - Volunteer Fire Brigade “Jarogniewice”, Zielona Gora, Poland (system implemented in September 2020), - Specialized Rescue Group of the Volunteer Fire Brigade, Glogow, Poland (system implemented in December 2020), - Bieszczady Group of the Mountain Volunteer Search and Rescue, Sanok, Poland (system to be implemented in spring 2021).

If applicable, choose the relevant working group (Ctrl touch to select more than one)

Please select the relevant item

Short description of the solution. Technical details if relevant. Keywords.

The SARUAV app works on a high-performance laptop. The user is provided with two software modules. The first one performs modelling of where a lost person might walk to, suggesting where a drone should operate. The second module aims to automatically detect persons in close-range aerial imagery acquired by consumer-grade RGB cameras installed onboard unmanned aerial vehicles. The detection is performed on a basis of two detectors which are methodically unlike each other.

The system works offline in the field, without any need for using external computational infrastructure. Data modelling and image processing is carried out quickly, i.e. the system processes 100 images within a few minutes. Apart from the dedicated GIS capabilities, it offers an easy-to-use verification panel which allows the user to make final decisions as to whether a given detected object is a missing person. In addition, the SARUAV mapping interface has been optimized following the eye-tracing research carried out among rescuers.

Keywords: unmanned aerial vehicle, person detection, search and rescue, image analysis, spatial analysis.

TRL of the proposed solution - Innovation stage (if applicable) 9: System Test, Launch & Operations

Web addresses/URL of flyers and information https://www.saruav.pl/

Video SARUAV_Eng_VO_Final_Version_Social_Media.mp4 (130 MB)

Expected/scheduled future developments

The system is ready to be implemented in Europe. Its future developments include the improvement in the detection performance through the development of detectors.

Generic comments

Each implementation of the SARUAV system is area-specific. It means that the SARUAV team does not only install the app itself, but the group prepares and upload dedicated data sets to be used in the process of modelling and detection of human. The company has developed IT tools that help to facilitate such an area-specific implementation process, and therefore the SARUAV team is ready for further deployments.