Detecting and Classifying Olive Leaf Pests and Diseases Using Optimal Deep Learning Techniques

Ali Elrowayati (1) , Yasir Swayeb (2) , Mohammed Baltu (3)
(1) Deptarment of Electronic Engineering, College of Industrial Technology, Misurata, Libya ,
(2) Faculty of Information Technology, Misurata University, Libya ,
(3) Faculty of Information Technology, Misurata University, Libya

Abstract

Agricultural advancements in technology and AI are transforming disease and pest control, boosting crop productivity. Classifying olive tree ailments is a challenge. Traditional methods are insufficient, requiring farmers and even experts to invest significant time and effort in manual identification. This paper explores CNNs for olive disease and pest classification. We experimented on two datasets: a local one from Libya with healthy leaves, olive fly, and jasmine moth (4,170 samples) and a public GitHub dataset with healthy leaves, rust mite blight, and peacock eye disease (6,961 samples). We compared pre-trained and untrained CNN models, finding the pre-trained Xception model achieved the highest local data accuracy (99%). Interestingly, the best-untrained model also excelled on local data (95%). The study further explored the impact of optimization algorithms (Adam and SGD), with Adam consistently achieving superior accuracy on both datasets.

References

Read More

Authors

Ali Elrowayati
elrowayati@yahoo.com (Primary Contact)
Yasir Swayeb
Mohammed Baltu
Elrowayati ع., Swayeb ي., & Baltu م. (2024). Detecting and Classifying Olive Leaf Pests and Diseases Using Optimal Deep Learning Techniques. Journal of Pure & Applied Sciences, 23(2), 167–177. https://doi.org/10.51984/jopas.v23i2.3441

Article Details

Capsule Network Implementation On FPGA

Salim A. Adrees , Ala A. Abdulrazeg
Abstract View : 325
Download :500