Periodic Reporting for period 1 - hivepoll (Polling beehives to monitor parasitic infestations and gather pollination information)
Période du rapport: 2019-06-01 au 2019-11-30
Due to challenges from climate change and an ever-growing world population the agricultural sector is under tremendous pressure to produce a lot more crop on limited acres. Bees play a vital role in save guarding a third of the world food production through their pollination activity.
With hivepoll we not only want to build a system which is able to protect beehives from parasites but establish bee colonies as a completely new kind of environmental and pollination data source. By introducing hivepoll we will be able to establish a novel data exchange platform between beekeepers and other stakeholder, e.g. in the agriculture market. Hivepoll offers a low-cost visual computing system to detect threats to beehives in real time but also unprecedented insights into the performance of bee colonies as pollinators and potentially how they are affected by adverse environmental conditions and climate change. By analyzing the amount and kind of polls collected as well as real-time flight characteristics we can turn beehives into novel biological sensors with far reaching consequences for the AgTech Market, beekeepers and public authorities. The data gathered by hivepoll allows to predict crop yield and using better models to manage crop failure risks, two features which are among the biggest opportunities for Big Data applications in agriculture. With hivepoll it will become possible to provide an early warning system in case of a low pollination performance as well as being able to counter these adverse effects and thus securing food production yields. Furthermore, an effective parasite detection will contribute to the health of bee colonies and their survival in good health over the winter which is a prime concern of bee keepers and farmers.
With hivepoll we not only want to build a system which is able to protect beehives from parasites but establish bee colonies as a completely new kind of environmental and pollination data source. By introducing hivepoll we will be able to establish a novel data exchange platform between beekeepers and other stakeholder, e.g. in the agriculture market. Hivepoll offers a low-cost visual computing system to detect threats to beehives in real time but also unprecedented insights into the performance of bee colonies as pollinators and potentially how they are affected by adverse environmental conditions and climate change. By analyzing the amount and kind of polls collected as well as real-time flight characteristics we can turn beehives into novel biological sensors with far reaching consequences for the AgTech Market, beekeepers and public authorities. The data gathered by hivepoll allows to predict crop yield and using better models to manage crop failure risks, two features which are among the biggest opportunities for Big Data applications in agriculture. With hivepoll it will become possible to provide an early warning system in case of a low pollination performance as well as being able to counter these adverse effects and thus securing food production yields. Furthermore, an effective parasite detection will contribute to the health of bee colonies and their survival in good health over the winter which is a prime concern of bee keepers and farmers.
In the SME Phase 1 project hivepoll concluded in 2019 we could validate the feasibility of poll analysis and quantification by means of computer vision algorithms applied on image data streams. In order to safeguard IPR a patent has been submitted to the Austrian office with plans to extend protection to EU and overseas. Furthermore a new hardware design for the hivepoll visual sensor has been developed. Finally, a thorough business plan modeling was conducted and a comprehensive and specific road to market strategy has been developed.
Based on our existing dataset of approx. 1600 hours of bee activity a subset with pollination information was generated. Compared to the only available public dataset the hivepoll dataset is more than 55 times bigger, totaling 1.912 images labelled as bees bearing pollen and 36.725 as bees without pollen. Using this new dataset a pollination detection and quantification algorithm have been developed based on machine learning methods. We could show, that poll detection with the hivepoll set-up is feasible and accurate. Based on two years of practical experience in the field an industrial product design study for the hivepoll hardware has been carried out as well including a number of innovations to increase bee detection and maintenance of the equipment.