@article {10.3844/ajeassp.2023.106.117,
article_type = {journal},
title = {Identification of the Presence of the "Swollen Shoot" Disease in Endemic Areas in Côte d'Ivoire Via Convolutional Neural Networks},
author = {Mamadou, Coulibaly and Kolo, Silue and Kouassi , Konan Hyacinthe and Asseu, Olivier},
volume = {16},
number = {3},
year = {2023},
month = {Oct},
pages = {106-117},
doi = {10.3844/ajeassp.2023.106.117},
url = {https://thescipub.com/abstract/ajeassp.2023.106.117},
abstract = {The detection of Swollen Shoot disease and its control is one of the major objectives of research related to sustainable cocoa farming in Côte d'Ivoire. To contain the epidemic, the Cocoa Coffee Council (CCC) in collaboration with the National Agency for Support to Rural Development (ANADER) is responsible for prospecting and delimiting the infected areas as well as for uprooting suspect cocoa plants. since there is currently no cure for this virus. However, this monitoring is done with the naked eye and mobilizes many human resources (planters and plant pathologists). This process is delicate and time-consuming, resulting in significant economic losses for both planters and Côte d’Ivoire. Convolutional Neural Networks (CNN) emerged from the study of the visual cortex of the brain. CNNs are particularly used in image processing and offer many applications related to precision agriculture. Over the past few years, thanks to the increase in computing power, and the amount of training data available, CNNs have been capable of superhuman performance on complex visual tasks. They are at the heart of automatic image and video classification systems. The objective of the work presented in this article is to establish a collaborative solution between CNN-based image processing and plant pathology. The solution will reduce the human labor time required by using algorithms to facilitate the identification of swollen shoot disease in a cocoa plantation. The use of images collected from a drone on cocoa plantations as input information, allowed our learning model, based on CNNs, to guide a new approach for automating Swollen diagnosis. Shoot with our model, we have achieved a level of accuracy of 98% based on the known symptoms.},
journal = {American Journal of Engineering and Applied Sciences},
publisher = {Science Publications}
}