Ide-cabe: chili varieties identification and classification system based leaf

Wiwin Suwarningsih, Purnomo Husnul Khotimah, Andri Fachrur Rozie, Andria Arisal, Dianadewi Riswantini, Ekasari Nugraheni, Devi Munandar, Rinda Kirana


Identifying good quality chili varieties can be done by observing their leaves. It is required for seed testing and certification processes. Currently, a manual leaf identification method is used in which human experts inspect a wide range of leaves every one to two months. An automatic method could increase the identification process. Deep learning has proven to be a prominent method for image classification. We investigate the performance of deep CNN models, as: AlexNet, VGG16, Inception-v3 and DenseNet-121; to classify chili variety. In this paper, we took images of leaves aged 10 days. A preprocessing strategy was taken to enrich the dataset and to boost the classification performance and to assess the proposed models’ quality. From this study, we acquired 12 classes of chili leaves dataset. We acquired performance accuracy ranging from 70.18% to 78.37%. Further, the classification results by DenseNet-121 obtained the highest accuracy of 78.37% and recall of 74.83%. The classifiers investigated in this study perform well despite the relatively small number of our dataset. These results encourage the application of such an approach in practice.


Chili variety; Deep CNN; Deep learning; Leaf identification; Variety classification

Full Text:




  • There are currently no refbacks.

Creative Commons License
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.

Bulletin of EEI Stats