[Paper Review] Image-Based Classification of Olive Species Specific to Turkiye with Deep Neural Networks

İrfan Atabaş, Hatice Karatas|arXiv (Cornell University)|Feb 26, 2026

Smart Agriculture and AI0 citations

TL;DR

The study uses stereo-imaged data and transfer-learned CNNs (MobileNetV2, EfficientNetB0) to classify five olive species in Turkiye, with EfficientNetB0 achieving 94.5% accuracy.

ABSTRACT

In this study, image processing and deep learning methodologies were employed to automatically classify local olive species cultivated in Turkiye. A stereo camera was utilized to capture images of five distinct olive species, which were then preprocessed to ensure their suitability for analysis. Convolutional Neural Network (CNN) architectures, specifically MobileNetV2 and EfficientNetB0, were employed for image classification. These models were optimized through a transfer learning approach. The training and testing results indicated that the EfficientNetB0 model exhibited the optimal performance, with an accuracy of 94.5%. The findings demonstrate that deep learning-based systems offer an effective solution for classifying olive species with high accuracy. The developed method has significant potential for application in areas such as automatic identification and quality control of agricultural products.

Motivation & Objective

• Motivate automatic identification of local olive species cultivated in Turkiye.
• Develop an image-based pipeline using deep learning for species classification.
• Evaluate multiple CNN architectures to identify the best-performing model.
• Assess the effectiveness of transfer learning in this agricultural imaging task.

Proposed method

• Capture images of five olive species using a stereo camera.
• Preprocess images to ensure suitability for CNN analysis.
• Train and compare MobileNetV2 and EfficientNetB0 architectures with transfer learning.
• Evaluate models on a held-out test set to determine accuracy.
• Identify the best-performing model based on accuracy and robustness.

Experimental results

Research questions

• RQ1 Can deep learning models accurately classify five olive species from Turkiye using image data?
• RQ2 Which CNN architecture (MobileNetV2 vs EfficientNetB0) yields higher classification accuracy for this task?
• RQ3 What is the achieved accuracy of the best-performing model on the test set?
• RQ4 Does transfer learning improve performance for this olive species classification problem?

Key findings

• EfficientNetB0 achieved the highest accuracy among the tested models, at 94.5%.
• Both MobileNetV2 and EfficientNetB0 were explored with transfer learning for this task.
• Stereo imaging was used to capture data suitable for CNN-based classification.
• Preprocessing steps ensured the images were appropriate for deep learning analysis.