Oktavian Abraham Lantang
Universitas Sam Ratulangi

Published : 2 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 2 Documents
Search

Komparasi Supervised dan Self-Supervised Learning untuk Klasifikasi Sel Darah pada Skenario Keterbatasan Data Berlabel Misella Mambu; Oktavian Abraham Lantang; Salaki Reynaldo Joshua
Riau Jurnal Teknik Informatika Vol. 5 No. 2 (2026): Juli 2026
Publisher : Universitas Pasir Pengaraian

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30606/rjti.v5i2.4555

Abstract

The development of blood cell image classification models is often hindered by the scarcity of labeled data. This study evaluates the effectiveness of supervised baseline models, supervised fine-tuning, and self-supervised learning (SSL) based on SimCLR and BYOL using the ResNet-50 and DenseNet-121 architectures. Data-limited simulation was performed using 10% labeled data for the classification stage and 90% unlabeled data for SSL pre-training on 17,092 images from 8 morphological classes. The results show that SSL achieves competitive performance, with the highest accuracy on ResNet-50 at 95.31% (SimCLR) and 94.27% (BYOL), and on DenseNet-121 at 93.75% (SimCLR) and 89.58% (BYOL). The SSL approach also demonstrated robustness against overfitting, good robustness against class imbalance, and Grad-CAM visualizations that align with medical expert assessments on ResNet-50. In conclusion, SSL effectively optimizes unlabeled data to improve the performance of classification models in scenarios with limited data annotation.
Analisis Kerapatan Pertahanan Sepak Bola Berbasis Computer Vision dan Transformasi Homografi Savio Hendriko Palendeng; Oktavian Abraham Lantang; Muhamad Dwisnanto Putro
Riau Jurnal Teknik Informatika Vol. 5 No. 3 (2026): November 2026
Publisher : Universitas Pasir Pengaraian

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30606/rjti.v5i3.4772

Abstract

Defensive compactness and inter-player distance are crucial tactical aspects in football. However, their evaluation has traditionally relied on subjective visual observation, which is highly susceptible to bias caused by perspective distortion from television broadcast cameras. This research aims to develop an automated data-driven spatiotemporal analytics system to quantitatively extract tactical metrics from single broadcast camera videos. This system provides an objective solution for coaching staff to precisely evaluate the effectiveness of defensive strategies, replacing manual observation methods. The proposed method integrates Computer Vision and Deep Learning, utilizing the YOLOv11m model for player detection, the BoTSORT tracking algorithm, and a color space filter for team differentiation. The 2D pixel coordinates are then converted into real-world pitch distance metrics using Homographic Transformation supported by Global Motion Estimation (GME) to overcome camera movement. Testing on static (Indonesia National Team) and dynamic (Bayern Munich) defensive scenarios showed that the detection model achieved an mAP50 precision of 0.986, with a tracking accuracy (MOTA) of 89.78%. Field distance projection validation recorded a very low error rate (MAPE) of 7.93% in the static calibration area. This computational system is proven reliable as a Sports Analytics instrument to reduce tactical interpretation bias, map Convex Hull and dyadic distances, and objectively support coaches strategic decision-making.