Garuda - Garba Rujukan Digital

Optimasi Deteksi Objek pada Video dengan Kompresi Region of Interest menggunakan Model YOLOv8 Assagaf, Azhryl; Muhtadi, Muis
Jurnal Pendidikan Informatika (EDUMATIC) Vol 9 No 2 (2025): Edumatic: Jurnal Pendidikan Informatika
Publisher : Universitas Hamzanwadi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.29408/edumatic.v9i2.30007

The demand for real-time object detection systems, such as those used in video surveillance and autonomous vehicles, drives the need for efficient data storage and transmission without compromising accuracy. One promising approach is Region of Interest (ROI)-based video compression, which preserves visual quality in important areas. This study aims to evaluate the impact of video compression on object detection accuracy using the YOLOv8 model through statistical analysis using Analysis of Variance (ANOVA), and to compare the effectiveness of uniform and ROI-based compression methods. Videos from the VIRAT Video Dataset were compressed using the Constant Rate Factor (CRF) parameter and evaluated based on mAP_50, mAP_50_95, and file size. ANOVA results indicate no statistically significant differences between the two methods. At CRF 50, file size can be reduced by over 60%, but mAP_50 accuracy drops below 50% due to quality degradation in non-ROI areas, which disrupts the spatial context required by the model. This study contributes by examining the compression tolerance limits of YOLOv8 and reveals that overall visual quality, rather than just object-focused quality, plays a crucial role in model performance. These findings have important implications for real-time applications such as CCTV and autonomous vehicles, where a balance between compression efficiency and detection accuracy is critical. Future studies may explore adaptive ROI approaches that consider dynamic object movement.

Sleeption: Aplikasi Deteksi Gerakan Video Real-time berbasis Mobile untuk Pemantauan Gangguan Tidur Bayi Andreanto, Dodik Dwi Andreanto; Muhtadi, Muis; Ariyandi, Haffas Zikri
Jurnal Pendidikan Informatika (EDUMATIC) Vol 9 No 2 (2025): Edumatic: Jurnal Pendidikan Informatika
Publisher : Universitas Hamzanwadi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.29408/edumatic.v9i2.30373

A baby's quality sleep is very important for his or her growth and development. Infant sleep disturbances have a negative impact on their physical health. Monitoring baby's sleep becomes very important to identify potential sleep disorders and ensure optimal sleep quality for babies. This research aims to build an android application called Sleeption using the waterfall model, a mobile-based baby sleep pattern monitoring application connected to a Raspberry Pi 3 device as an edge computing device. The development phase includes the stages of analysing the needs of parents in monitoring activities, application design, implementation of application design, verification of the success of features in the application, application and application testing along with periodic maintenance on the application. Data collection is done by observation with the development team on the needs of parents in conducting baby sleep monitoring activities. The result of our findings is a mobile application that helps parents in the process of monitoring their baby's sleep activities. The results of black box testing show that the Sleeption application is able to provide accurate and real-time information with a latency of 41 ms about the baby's sleep activity. The application of the Sleeption application helps parents in monitoring their baby's sleep patterns, along with the automation of reports on changes in baby sleep patterns helping parents to identify potential sleep disorders more accurately.

Metode Frame Difference untuk Deteksi Gerakan Tidur Bayi berbasis Computer Vision Ariyandi, Haffas Zikri; Muhtadi, Muis; Andreanto, Dodik Dwi
Jurnal Pendidikan Informatika (EDUMATIC) Vol 9 No 1 (2025): Edumatic: Jurnal Pendidikan Informatika
Publisher : Universitas Hamzanwadi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.29408/edumatic.v9i1.29004

Monitoring a baby's sleep is a critical task for parents, especially when balancing household responsibilities. This study combines the MobileNet-SSD object detection model with the Frame Difference method to analyze sleep movements based on motion thresholds. The system's performance was evaluated by calculating accuracy, precision, recall, and latency, implemented on both laptop and Raspberry Pi devices, and tested using 720p and 480p resolution videos. Results showed accuracy of 82%, precision of 81%, and recall of 92% at 720p, and accuracy of 77%, precision of 80%, and recall of 86% at 480p. However, the Raspberry Pi exhibited a latency of 400ms, 10 times higher than the laptop's 41.28ms latency. Compared to optical flow, this method offers ease of use, and lower computational complexity. The results of this study highlight the impact of resolution on motion detection accuracy, where higher-resolution videos yield more optimal performance. Limitations under low-light conditions suggest potential improvements using deep learning techniques like YOLO and Mediapipe to detect eye conditions. This research contributes to the development of computer vision where the frame differential and object detection methods are proven to provide a fairly high level of accuracy in detecting movement.

Risk Detection System for Children Putting Objects into Mouth Based on Computer Vision using YOLOv11n: Sistem Deteksi Risiko Anak Memasukkan Benda ke dalam Mulut Berbasis Computer Vision menggunakan YOLOv11n Ramadan, Nofri; Muhtadi, Muis; Rafi, Muhammad Zulfiqar; Waluyo, Rinakit Estu
Indonesian Journal of Innovation Studies Vol. 26 No. 4 (2025): October
Publisher : Universitas Muhammadiyah Sidoarjo

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.21070/ijins.v26i4.1851

Background: Young children commonly place objects near or into their mouths, creating safety concerns that require constant supervision. Specific Background: Advances in computer vision enable real-time recognition of hands, faces, and objects, allowing automated identification of behaviors that may lead to hazardous mouth-related interactions. Knowledge Gap: Few systems combine object detection and human pose estimation specifically to assess risks related to toddler hand–mouth–object interactions in real-time environments. Aims: This study develops a risk detection system using YOLOv11n to recognize hands, faces, and objects while classifying conditions into safe or risky based on Euclidean distance between hand and mouth keypoints. Results: The system produces 92% accuracy in scenarios without objects and 74% in scenarios with objects, demonstrating its capability to differentiate between safe and risky conditions. Novelty: This research introduces an integrated spatial analysis approach that evaluates real-time proximity among hands, mouth, and objects, rather than detecting these elements independently. Implication: The system provides practical potential for real-time child safety monitoring, offering earlier awareness of mouth-related object risks in various activity settings. Highlights • The system identifies hand–mouth proximity to classify safe and risky situations.• Integrated pose estimation and object detection enable spatial risk assessment in real time.• The model supports early awareness of mouth-related object risks during toddler activities. Keywords Object Detection, Pose Estimation, Computer Vision, YOLO, Child Safety Monitoring

Application of YOLOv8 in Computer Vision-Based Facial Expression Detection of Toddlers: Penerapan YOLOv8 dalam Deteksi Ekspresi Wajah Balita Berbasis Computer Vision Waluyo, Rinakit Estu; Muhtadi, Muis; Ramadan, Nofri; Rafi, Muhammad Zulfiqar
Indonesian Journal of Innovation Studies Vol. 26 No. 4 (2025): October
Publisher : Universitas Muhammadiyah Sidoarjo

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.21070/ijins.v26i4.1853

Background: Facial expressions are a primary form of nonverbal communication for toddlers whose verbal abilities are still developing. Specific background: Advances in computer vision and deep learning have enabled real-time facial expression detection; however, most existing systems are designed using adult facial datasets and pretrained models. Knowledge gap: Research focusing on toddler facial expression detection using models trained exclusively on toddler data without pretrained weights remains limited. Aims: This study applies YOLOv8 to detect happy, sad, and neutral facial expressions of toddlers in real time using a model trained from scratch. Results: The proposed system achieved an average detection accuracy of 86%, with precision of 0.944, recall of 0.933, and mean Average Precision at 0.5 of 0.966, demonstrating stable real-time performance under varying lighting conditions. Novelty: The study demonstrates that YOLOv8 can learn toddler-specific facial expression patterns without relying on pretrained weights derived from adult facial data. Implications: The findings indicate the feasibility of deploying real-time toddler facial expression detection systems to support emotional monitoring in childcare and early education environments. Highlights Real-time detection of toddler facial expressions using YOLOv8. Model training conducted entirely from scratch using toddler facial datasets. Consistent detection performance observed in natural and varied environments. Keywords Facial Expression Detection, Toddler, YOLOv8, Computer Vision, Real-Time System

Evaluation of YOLOv8n, YOLOv10n, and YOLOv11n for Early Detection of Seizures in Infants: Evaluasi YOLOv8n, YOLOv10n, dan YOLOv11n untuk Deteksi Dini Kejang pada Bayi Rafi, Muhammad Zulfiqar; Muhtadi, Muis; Ramadan, Nofri; Waluyo, Rinakit Estu
Indonesian Journal of Innovation Studies Vol. 27 No. 1 (2026): January
Publisher : Universitas Muhammadiyah Sidoarjo

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.21070/ijins.v27i1.1855

Background: Infant epilepsy is a serious neurological condition that is difficult to recognize visually because seizure manifestations often resemble normal infant reflexes. Specific background: Advances in computer vision and deep learning have enabled image-based seizure detection as a non-invasive approach for early identification in infants. Knowledge gap: Comparative evaluations of different YOLO model generations under identical experimental settings for infant seizure detection remain limited. Aims: This study evaluates and compares the performance of YOLOv8n, YOLOv10n, and YOLOv11n in detecting seizure activity from infant image data. Results: Using a dataset of 645 images across seizure and neutral classes, YOLOv11n achieved the highest precision, recall, and mean average precision at stricter localization thresholds, while YOLOv10n demonstrated the fastest inference time. Novelty: This study provides a systematic cross-generation evaluation of YOLO nano variants for infant seizure detection using uniform training parameters. Implications: The findings support informed selection of YOLO architectures for non-invasive, image-based early seizure detection systems in clinical and real-time monitoring contexts. Highlights • YOLOv11n shows the most stable detection performance for infant seizure images• YOLOv10n achieves the fastest inference time under identical training settings• Cross-version evaluation reveals clear trade-offs between detection stability and processing speed Keywords YOLO, Epilepsy, Computer Vision, Object Detection, Infant Seizure Detection

Title

Found 6 Documents
Search

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Title Search

Found 6 Documents Search

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Title

Found 6 Documents
Search