Mots-clé
(Allemand)
|
Kirschessigfliege, Beeren, Drosophila suzukii, Kirsche, Erdbeere, Himbeere, Brombeere, Blaubeere, Holunderbeere, automatisierte Überwachung, klebrige Fallen, Drohne, Computervision, Entscheidungshilfesystem
|
Mots-clé
(Anglais)
|
Fruitflies, berries, Drosophila suzukii, cherry, strawberry, raspberry, blackberry, blueberry, elderberry, Automated monitoring, sticky traps, drone, computer vision, decision support system
|
Mots-clé
(Français)
|
drosophile japonaise, baies, Drosophila suzukii, cerise, fraise, framboise, mûre, myrtille, sureau, surveillance automatisée, pièges collants, drone, vision par ordinateur, système d'aide à la décision
|
Description succincte
(Anglais)
|
Drosophila suzukii has become a serious pest in Europe since its spread in 2008 to Spain and Italy, attacking many soft-skinned crops such as several berry species, cherry and grapevines. Pest monitoring is the basis of its control. Therefore, an efficient and accurate monitoring system is essential in order to identify the presence of D. suzukii in the crops and the surrounding area, and to prevent damage to economically valuable fruit crops. Existing methods for monitoring D. suzukii are costly, time and labor intensive and consequently conducted at low spatial resolution and prone to errors. The project therefore propose to develop a novel system to overcome current monitoring limitations consisting of traps which are monitored by means of an Unmanned Aerial Vehicle (UAV) and an automatic image processing pipeline for the identification and count of number of D. suzukii per trap location. The automated monitoring has an advantage over current methods in terms of (1) labor intensity, (2) sampling interval, (3) automatic integration into DSS, (4) monitoring of diverse and even hardly accessible habitats, and (5) population monitoring in vast areas in relation to climatic and other geo-processed parameters. A multi-variable sticky trap evaluation will allow selecting the most suitable one to attract the target insect. A small multi-rotor UAV platform will be flown at multiple intervals to capture high resolution color aerial photographs of the insect traps. The photographs will be subjected to image processing algorithms to identify the presence or absence of D. suzukii and their counts. The data collected will be transferred to a decision support system (DSS) to provide valuable information for growers in a format that is both meaningful and accessible, thereby demonstrating the added value and social importance of applied science and technology to the wider community and food security.
|
Publications / Résultats
(Anglais)
|
The AAPM project developed knowledge on trapping spotted wing drosophila (SWD) on planar surfaces such as sticky traps and the most suitable attractant trap colour. Since commercially available sticky traps are not suitable for field application due to low catch rate, ZHAW developed a new photographable trap baited with liquid lure as an additional attractant. WUR learned how to automatically count the target insects on ground based and airborne high-resolution imagery of the traps using deep learning methodology. In natural and simulated environments UoA tested multiple Unmanned Aerial Vehicles (UAV) platforms for their autonomous flight capacity and sensor optical resolution. The consortium collected airborne imagery, and used the trained algorithm to successfully discriminate SWD from bycatch and count the target insect on these images. Feasibility of autonomous flights with were investigated and tested.
|
