Plant phenology recognition using deep learning: Deep-Pheno

Abstract

Monitoring phenology of agricultural plants is a critical understanding in precision agriculture. Vital improvements can be achieved with precise detection of phenological change of plants which would henceforth improve the timing for the harvest, pest control, yield prediction, farm monitoring, disaster warning etc. Many countries across the world have been developing initiatives to build national agriculture monitoring network systems, since inferring the phenological information contributes to a better understanding of relationships between productivity, vegetation health and environmental conditions. In this paper, we utilize a deep learning architecture to recognize and classify phenological stages of several types of plants purely based on the visual data captured every half an hour by cameras mounted on the ground agro-stations that have been planted all over Turkey as part of an agriculture monitoring network system. A pre-trained Convolutional Neural Network architecture (CNN) is employed to automatically extract the features of images. In order to evaluate the performance of the approach proposed in this paper, the results obtained through CNN model are compared with those obtained by employing hand crafted feature descriptors. Experimental results suggest that CNN architecture outperforms the machine learning algorithms based on hand crafted features for the discrimination of phenological stages.