Detection of Milkfish Freshness Levels Using YOLOv11 Nano Based on Fish Eye Image Analysis

Putri, Prudencia Belva Cynara Trana (2026) Detection of Milkfish Freshness Levels Using YOLOv11 Nano Based on Fish Eye Image Analysis. Undergraduate thesis, UPN Veteran Jawa Timur.

	Text (Cover) 1522081010307.-cover.pdf.pdf Download (2MB)
	Text (Bab 1) 1522081010307.-bab1.pdf.pdf Download (107kB)
	Text (Bab 2) 1522081010307.-bab2.pdf.pdf Restricted to Repository staff only until 21 July 2029. Download (4MB) \| Request a copy
	Text (Bab 3) 1522081010307.-bab3.pdf.pdf Restricted to Repository staff only until 21 July 2029. Download (21MB) \| Request a copy
	Text (Bab 4) 1522081010307.-bab4.pdf.pdf Restricted to Repository staff only until 21 July 2029. Download (8MB) \| Request a copy
	Text (Bab 5) 1522081010307.-bab5.pdf.pdf Download (90kB)
	Text (Daftar Pustaka) 1522081010307.-daftarpustaka.pdf.pdf Download (140kB)
	Text (Lampiran) 1522081010307.-lampiran.pdf.pdf Restricted to Repository staff only until 21 July 2029. Download (570kB) \| Request a copy

Abstract

This research aims to develop a detection and classification system for the freshness level of milkfish based on YOLOv11 Nano using fish eye images on Android devices. The research methods used include data collection stages, data preprocessing consisting of data selection and cleaning, image standardization and renaming, object annotation, dataset splitting, resolution resizing, augmentation, and normalization, followed by model training, model evaluation, application design, application implementation, and application testing. During the training phase, several parameter scenarios were tested, namely dataset splitting, learning rate, and the number of epochs to obtain the best-performing model. The research results show that the best scenario was obtained in the second scenario with a dataset split of 80%:10%:10%, a learning rate of 0.01, and 100 epochs. The configuration resulted in a Precision value of 94.3%, Recall of 91.1%, mAP@50 of 90.6%, mAP@50-95 of 58.7%, and an F1-Score of 92.7% on the test data. The best model was then converted to TensorFlow Lite format and implemented in an Android application. The application testing results show an average inference time of 288.83 ms per image with 3.46 FPS using the GPU and 785 ms per image with 1.27 FPS using the CPU, where the GPU usage is 2.72 times faster than the CPU. Additionally, the application has efficient resource usage with an average CPU usage of 14.9% and a memory usage of 187.51 MB. Based on these results, the research demonstrates that YOLOv11 Nano is capable of accurately and efficiently detecting and classifying the freshness level of milkfish on Android devices.

Item Type:

Thesis (Undergraduate)

Contributors:

Contribution	Contributors	NIDN/NIDK	Email
Thesis advisor	Al Haromainy, M. Muharrom	0701069503	muhammad.muharrom.if@upnjatim.ac.id
Thesis advisor	Anggraeny, Fetty Tri	0711028201	fettyanggraeny.if@upnjatim.ac.id

Uncontrolled Keywords:

Milkfish Freshness, Fish Eye Images, YOLOv11 Nano, Object Detection, Android

Subjects:

Q Science > Q Science (General)
Q Science > QA Mathematics > QA76 Computer software
Q Science > QA Mathematics > QA76.87 Neural computers

Divisions:

Faculty of Computer Science > Departemen of Informatics

Depositing User:

Prudencia Belva Cynara Trana Putri

Date Deposited:

22 Jul 2026 04:54

Last Modified:

22 Jul 2026 06:52