Elmayati
Universitas Bina Insan, Lubuklinggau

Published : 1 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 1 Documents
Search

KLASIFIKASI JENIS DAN STADIA PERTUMBUHAN JANGKRIK DARI PETERNAKAN MENGGUNAKAN EFFICIENTNETV2 DAN XGBOOST Yogi Irawan; Muhammad Akbar; Elmayati; Novi Lestari
JUTIM (Jurnal Teknik Informatika Musirawas) Vol. 11 No. 3 (2026): JUTIM (Jurnal Teknik Informatika Musirawas)
Publisher : LPPM UNIVERSITAS BINA INSAN

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.32767/jutim.v11i3.3315

Abstract

The identification of cricket species and growth stages in farm environments is still performed manually, making the process prone to subjectivity and inefficiency. This study aims to develop an automated classification system based on digital images using a hybrid EfficientNetV2-B1 and XGBoost approach. The dataset consists of 2,400 images across six classes, representing combinations of two cricket types, madu and seliring, and three growth stages: early nymph, middle nymph, and imago. The data were divided into 1,920 training images, 240 validation images, and 240 test images at a ratio of 80:10:10. A data leakage audit using SHA-256 and perceptual hash (pHash) found no identical duplicate images or image pairs meeting the similarity threshold. The experiments compared the hybrid model without HSV, the hybrid model with HSV color features, and the Softmax baseline model. Evaluation on 240 test images showed that the hybrid model without HSV achieved an accuracy of 92.92%, precision of 93.21%, recall of 92.92%, and F1-score of 92.94%. The model with HSV features achieved an accuracy of 90.00%, while the Softmax baseline achieved an accuracy of 88.33% and an F1-score of 88.48%. The best-performing hybrid model was implemented in a Streamlit-based web application. Future research should increase dataset diversity and image quality, as well as implement the model operationally in farm environments to evaluate its performance under real-world conditions.