![](/static/ps_publications/img/article-icon.png)
Ozone nano-mist disinfection of insect pests and automatic spraying system based on deep learning technology
Artykuł w czasopiśmie
MNiSW
20
Lista 2024
Status: | |
Autorzy: | Ebihara Kenji, Mitsugi Fumiaki, Aoqui Shin-ichi, Boiko Oleksandr, Stryczewska Henryka, Yamashita Yoshitaka, Baba Seiji |
Dyscypliny: | |
Aby zobaczyć szczegóły należy się zalogować. | |
Rok wydania: | 2024 |
Wersja dokumentu: | Drukowana | Elektroniczna |
Język: | angielski |
Numer czasopisma: | 2 |
Wolumen/Tom: | 18 |
Numer artykułu: | e02003 |
Strony: | 1 - 13 |
Scopus® Cytowania: | 0 |
Bazy: | Scopus | Google Scholar |
Efekt badań statutowych | NIE |
Finansowanie: | This work was funded by the Polish National Agency of Academic Exchange (NAWA) under the project “POL-JAP ENERGO-ECO”, Grant PPI_APM_2019_1_00009 as well as Lublin University of Technology Grants FD-20/EE-2/416, FD-20/EE-2/401 and 5/GnG/2022. |
Materiał konferencyjny: | NIE |
Publikacja OA: | TAK |
Licencja: | |
Sposób udostępnienia: | Otwarte czasopismo |
Wersja tekstu: | Ostateczna wersja opublikowana |
Czas opublikowania: | W momencie opublikowania |
Data opublikowania w OA: | 1 czerwca 2024 |
Abstrakty: | angielski |
We propose an ozone nano-mist spraying system which is composed of an ozone nano-mist generator and an automatic spraying system in agriculture. The highly dense ozone generated by dielectric barrier discharges is injected into water nano-mist flow ejected from an ultrasonic humidifier(1.7Mhz). This disinfection system is controlled to spray in real time the ozone nano-mist on the detected pests in response to the treatment conditions given by the deep learning technology. Six species of insect pests (aphid, moth, beetle, fly, whitefly and ant) were selected to study control performance of the disinfection spraying on these pests in the greenhouse. The disinfection rates of the 6 species of insect pests were measured in advance at various nano-mist conditions. The YOLO object detection architecture based on deep learning is adopted to acquire information about insect species and their number of the target insect pests appeared on the photo images. The automatic nano-mist spraying system is operated by the signals from the Raspberry Pi which communicates remotely with the main computer using Wi-Fi. In these processes of the training and validation of the pest dataset, mean average precision values of 96.8% (mAP@0.5) and 70.1% (mAP@0.5:0.95) were achieved for all classes(species). The computing time for this training and the validation was 6 minutes. When the object targets are winged insect pests such as moths, flies and whiteflies, the testing time was 11-12 sec and the consuming time for the following spraying procedure was 30 sec at an ozone flow rate of 110 g-O3 m-3. The disinfection rate at this spraying process was near 100% and closed to the rate of the conventional chemical pesticide treatment. The results suggest that the proposed ozone nano-mist disinfection system can be used practically to disinfect remotely winged insect pests in greenhouses. |