Skripsi

KLASIFIKASI JENIS BAKTERI CITRA MIKROSKOPIS MENGGUNAKAN CONVOLUTIONAL NEURAL NETWORK DENGAN ARSITEKTUR CONVNEXT DAN RESNET-50

---

Identifying bacteria traditionally takes 24–72 hours and is prone to human error. With bacterial infections and antimicrobial resistance causing over 1 million deaths annually, faster and more accurate methods are urgently needed. This study compares two deep learning models, ResNet-50 (a classic CNN) and ConvNeXt-Tiny (a modern CNN) for classifying microscopic bacteria images using the DIBaS dataset, which contains 2,033 images across 33 bacterial species. Four model configurations were tested with varying learning rates and batch sizes, using transfer learning and data augmentation to improve performance. ConvNeXt-Tiny achieved the best results with 99% accuracy, precision, recall, and F1-score. ResNet-50 followed closely at 98.7% accuracy. Batch size 32 and learning rate 0.001 consistently produced better outcomes. The best model also proved robust under Test Time Augmentation, maintaining 99.1% accuracy. A web-based application was built using Streamlit to make the classification tool accessible to users. This research offers practical guidance for selecting CNN architectures in medical and biological image analysis.

---

Availability

Detail Information

Series Title

-

Call Number

T1955792026

Publisher

Indralaya : Prodi Teknik Informatika, Fakultas Ilmu Komputer Universitas Sriwijaya., 2026

Collation

xvi, 108 hlm.; ilus.; 29 cm

Language

Indonesia

ISBN/ISSN

-

Classification

006.370 7

Content Type

Text

Media Type

-

Carrier Type

-

Edition

-

Subject(s)

Klasifikasi Citra
Prodi Teknik Informatika

Specific Detail Info

-

Statement of Responsibility

TUTI

Other version/related

File Attachment

Comments

---

You must be logged in to post a comment