Grapevine flower estimation on 2D images taken under field conditions and with stochastic scene
- Aquino, A.
- Millán, B.
- Diago, M.P.
- Rey, C.
- Fernández, J.
- Gutiérrez, S.
- Tardáguila, J.
Year of publication: 2015
Pages: 3
Congress: International Computer Vision Summer School. 12-18 July. (2015. Sicily, Italy)
Type: Conference Poster
Abstract
A great challenge for the scientific community in viticulture is early yield prediction. For this, flowering and fruit set assessment are of special interest since they highlyinfluence grapevine yield. A new methodology based on mathematical morphology and pyramidal decomposition for segmenting inflorescence grapevine flowers in 2D digital images takenunder field conditions is presented. The algorithm was tested on 40 images of 4 different Vitisvinifera L. varieties, providing Precision and Recall values of 83.38% and 85.01%, respectively.