Year: 2025
Type: article
Abstract: This paper presents the use of deep learning network-based Convolutional Neural Networks (CNNs) to enhance the efficiency of classifying purebred Thai native chickens for conservation purposes. This study specifically focuses on Thai native chicken species known as Leung Hang Khao. Due to the significant genetic diversity of the Thai native chickens, it typically requires experts to accurately identify the breeds. There are four groups of the Thai native chickens that were considered in this work; namely, purebred Leung Hang Khao male, purebred Leung Hang Khao female, crossbred male, and crossbred female. A total of 1,000 images have been collected, in which 250 images are from each group. Then, the data is divided into three sets which are training set, validation set, and testing set, in the ratios of 60:20:20, 70:20:10, and 80:10:10, respectively. Four architectures of the CNNs have been employed for verification, i.e., LeNet-5, CNN1, CNN2, and CNN3, with epochs set at 10, 20, and 50 epochs for each architecture. The results show that the CNN1 architecture with an 80:10:10 ratio and 10 epochs yielded the highest accuracy in learning, validation, testing, and prediction. Moreover, it required relatively less testing time with predicted accurate results of 100 %. The obtained results demonstrate that using the deep learning network-based convolutional neural network with a simple architecture setting can effectively classify Thai native chicken breeds.
Institution: Rajamangala University of Technology Tawan-ok
Language: Thai
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
SDG: Zero hunger
Open Access status: diamond
OpenAlex ID: W7215605515