Page 89 - IJAMD-2-2

International Journal of AI for
Materials and Design Fruit image detection using AI

A

B C

Figure 2. Automated fruit sorting system developed in this study.
(A) The conceptual drawing of the system. (B) The fabricated side view of
the system. (C) The fabricated front view of the system. Image produced
by the authors.

Figure 3. Complete experimental setup of the system. Image produced Figure 4. An overall framework of the method for the automated fruit
by the authors. sorting system.
Abbreviation: SVM: Support vector machine.
that combined features provide more comprehensive Table 1. Summary of the experimental results for fruit
information about the fruit characteristics, leading to classification models based on image features
improved prediction outcomes.
Feature Accuracy (%) Precision (%) F1 score (%)
The results were also favorable in terms of precision,
which measures the accuracy of positive predictions. In Color feature 85.0 82.0 81.0
this study, the model achieved a precision score of 88%, Shape feature 78.0 75.0 72.5
indicating high reliability in its positive classifications. Texture feature 80.0 78.0 76.5
The accuracy is particularly important in agricultural Size feature 76.0 74.0 73.0
applications, where precise identification of fruit types can Combined features 90.0 88.0 86.5
significantly influence sorting efficiency and marketing
decisions.
combined features, the model achieved a maximum recall
Recall (or sensitivity) measures the effectiveness of a score of 85%, efficiently capturing most of the true positive
model in identifying relevant instances. In this study, using cases. This measure is important in ensuring that no fruit

Volume 2 Issue 2 (2025) 83 doi: 10.36922/IJAMD025150011