In Italy, peaches are paid according to size, color and appearance (i.e., no external damage). Having information related to these parameters directly at harvest could support growers and the fruit chain in improving fruit quality for consumers as well as to increase grower income. In this study, a computer vision system was developed which aims to count and size peach fruit in bins in the field, at harvest. Two different depth cameras (Intel RealsenseD435i and D455) and two different light conditions (natural and artificial) were tested, to assess potential problems (e.g., interference from natural light in fruit detection) and to evaluate the best system set-up for future developments. While automated fruit identification appears less problematic, the system has largely overestimated fruit size in all the conditions tested. The D435i camera with artificial lighting obtained the best results with a root means square error (RMSE) of 16.7 mm, compared to the reference fruit diameter. The results obtained are however promising and suggest solutions to further improve the system. Future work will focus on improving the system for sizing, color estimation (color intensity and extension) and georeferentiation of data directly in the field. The idea is to develop a low-cost plugin for harvesting platforms that can support growers, and the peach chain, to start connecting all post-harvest operations to pre-harvest conditions, and to fruit quality at time of harvest.

Pilot study of a computer vision system for in-field peach fruit quality evaluation

Bortolotti, G.
Primo
;
Piani, M.
Secondo
;
Mengoli, D.;Corelli Grappadelli, L.
Penultimo
;
Manfrini, L.
Ultimo
2022

Abstract

In Italy, peaches are paid according to size, color and appearance (i.e., no external damage). Having information related to these parameters directly at harvest could support growers and the fruit chain in improving fruit quality for consumers as well as to increase grower income. In this study, a computer vision system was developed which aims to count and size peach fruit in bins in the field, at harvest. Two different depth cameras (Intel RealsenseD435i and D455) and two different light conditions (natural and artificial) were tested, to assess potential problems (e.g., interference from natural light in fruit detection) and to evaluate the best system set-up for future developments. While automated fruit identification appears less problematic, the system has largely overestimated fruit size in all the conditions tested. The D435i camera with artificial lighting obtained the best results with a root means square error (RMSE) of 16.7 mm, compared to the reference fruit diameter. The results obtained are however promising and suggest solutions to further improve the system. Future work will focus on improving the system for sizing, color estimation (color intensity and extension) and georeferentiation of data directly in the field. The idea is to develop a low-cost plugin for harvesting platforms that can support growers, and the peach chain, to start connecting all post-harvest operations to pre-harvest conditions, and to fruit quality at time of harvest.
2022
Acta Horticulturae
315
322
Bortolotti, G.; Piani, M.; Mengoli, D.; Corelli Grappadelli, L.; Manfrini, L.
File in questo prodotto:
Eventuali allegati, non sono esposti

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11585/913154
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? 0
social impact