Comparison CNN and Mobilenet_v2 Model for Oil Palm FBB Ripeness Classification


  • Muhammad Hanafi Mohtar Luddin student
  • Munirah Ab Rahman UTHM


Deep Learning, CNN, Mobilenet_v2, Classification


In 2021, the value of exports of palm oil and derivatives increased by 40% to RM91.4 billion. This makes the palm oil industry the fourth largest contributor to the national economy and palm oil as well as the largest contributor to Malaysia's commodity exports. The selection of the ripeness of fresh fruit bunches (FFB) of oil palm is very important to obtain the best quality palm oil. But the selection process is still carried out manually, while the selection process through machines is still poorly introduced in the palm oil industry. To overcome this problem, this study has produced a deep learning system using the CNN and Mobilenet_v2 model algorithms. The data set used consists of 1542 images of oil palm FFB for both the training and validation process. The CNN model was built as the primary model for this study, while the Mobilenet_v2 model was built to differentiate accuracy performance. Based on the experiments conducted in this study, the CNN Model managed to classify the new image of oil palm FFB with a percentage of 99.99% accuracy based on epoch = 30 compared to the Mobilenet_v2 model with a percentage of 14.89% accuracy. Finally, the study has proven that the CNN model successfully classifies the 10 categories of oil palm FFB ripeness using data set of 1542 images.




How to Cite

Mohtar Luddin, M. H., & Ab Rahman, M. (2022). Comparison CNN and Mobilenet_v2 Model for Oil Palm FBB Ripeness Classification. Evolution in Electrical and Electronic Engineering, 3(2), 146–154. Retrieved from



Computer and Network