Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Mirna Yunita has a total of 36 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    YOLO-LIO: A Real-Time Enhanced Detection and Integrated Traffic Monitoring System for Road Vehicles

    by Rachmat Muwardi, Haiyang Zhang, Hongmin Gao, Mirna Yunita, Rizky Rahmatullah, Ahmad Musyafa, Galang Persada Nurani Hakim, Dedik Romahadi

    Traffic violations and road accidents remain significant challenges in developing safe and efficient transportation systems. Despite technological advancements, improving vehicle detection accuracy and enabling real-time traffic management remain critical research priorities. This study proposes YOLO-LIO, an enhanced vehicle detection framework designed to address these challenges by improving small-object detection and optimizing real-time deployment. The system introduces multi-scale detection, virtual zone filtering, and efficient preprocessing techniques, including grayscale transformation, Laplacian variance calculation, and median filtering to reduce computational complexity while maintaining high performance. YOLO-LIO was rigorously evaluated on five datasets, GRAM Road-Traffic Monitoring (99.55% accuracy), MAVD-Traffic (99.02%), UA-DETRAC (65.14%), KITTI (94.21%), and an Author Dataset (99.45%), consistently demonstrating superior detection capabilities across diverse traffic scenarios. Additional system features include vehicle counting using a dual-line detection strategy within a virtual zone and speed detection based on frame displacement and camera calibration. These enhancements enable the system to monitor traffic flow and vehicle speeds with high accuracy. YOLO-LIO was successfully deployed on Jetson Nano, a compact, energy-efficient hardware platform, proving its suitability for real-time, low-power embedded applications. The proposed system offers an accurate, scalable, and computationally efficient solution, advancing intelligent transportation systems and improving traffic safety management.


  • Open Access

    Optimization of YOLOv4-Tiny Algorithm for Vehicle Detection and Vehicle Count Detection Embedded System

    by Rachmat Muwardi, Ivan Prasetyo Nugroho, Ketty Siti Salamah, Mirna Yunita, Rizky Rahmatullah, Gregorius Justin Chung

    Currently, the implementation of object detection systems in the traffic sector is minimal. CCTV cameras on highways and toll roads are primarily used to monitor traffic conditions and document violations. However, the data recorded by these cameras can be further utilized to enhance traffic management systems. The author proposes a vehicle detection and counting system using YOLOv4-Tiny. The research aims to improve vehicle detection and counting accuracy by employing a median filter and grayscale processing, which simplify object detection. The proposed YOLOv4-Tiny algorithm has shown impressive results on various datasets, including MAVD, GRAM-RTM, and author dataset. The system achieved a detection accuracy of 98.95% on the MAVD dataset, 99.5% on the GRAM-RTM dataset (comparable to YOLOv4), and 99.1% on the author dataset. Furthermore, the system operates at 25 frames per second (FPS), a notably high rate compared to other methods. While the system demonstrates excellent accuracy in counting cars, it encounters some accuracy loss with other vehicle classifications. The author concludes that the system is highly suitable for real-world applications but notes that inaccurate labeling can lead to vehicle counting errors.


  • Open Access

    L2T-DLN: Learning to Teach with Dynamic Loss Network

    by Zhoyang Hai, Liyuan Pan, Xiabi Liu, Zhengzheng Liu, Mirna Yunita

    With the concept of teaching being introduced to the machine learning community, a teacher model start using dynamic loss functions to teach the training of a student model. The dynamic intends to set adaptive loss functions to different phases of student model learning. In existing works, the teacher model 1) merely determines the loss function based on the present states of the student model, i.e., disregards the experience of the teacher; 2) only utilizes the states of the student model, e.g., training iteration number and loss/accuracy from training/validation sets, while ignoring the states of the loss function. In this paper, we first formulate the loss adjustment as a temporal task by designing a teacher model with memory units, and, therefore, enables the student learning to be guided by the experience of the teacher model. Then, with a dynamic loss network, we can additionally use the states of the loss to assist the teacher learning in enhancing the interactions between the teacher and the student model. Extensive experiments demonstrate our approach can enhance student learning and improve the performance of various deep models on real-world tasks, including classification, objective detection, and semantic segmentation scenarios.

  • Open Access

    Design Human Object Detection Yolov4-Tiny Algorithm on ARM Cortex-A72 and A53

    by Rachmat Muwardi, Ahmad Faizin, Puput Dani Prasetyo Adi, Rizky Rahmatullah, Yanxi Wang, Mirna Yunita, Dendi Mahabror

    Currently, many object detection systems still use devices with large sizes, such as using PCs, as supporting devices, for object detection. This makes these devices challenging to use as a security system in public facilities based on human object detection. In contrast, many Mini PCs currently use ARM processors with high specifications. In this research, to detect human objects will use the Mini PC Nanopi M4V2 device that has a speed in processing with the support of CPU Dual-Core Cortex-A72 (up to 2.0 GHz) + Cortex A53 (Up to 2.0 GHz) and 4 Gb DDR4 Ram. In addition, for the human object detection system, the author uses the You Only Look Once (YOLO) method with the YoloV4-Tiny type, With these specifications and methods, the detection rate and FPS score are seen which are the feasibility values for use in detecting human objects. The simulation for human object recognition was carried out using recorded video, simulation obtained a detection rate of 0,9845 or 98% with FPS score of 3.81-5.55. These results are the best when compared with the YOLOV4 and YOLOV5 models. With these results, it can be applied in various human detection applications and of course robustness testing is needed.


  • Open Access

    Fast Human Recognition System on Real-Time Camera

    by Yuliza Yuliza, Rachmat Muwardi, Mustain Rhozaly, Lenni Lenni, Mirna Yunita, Galatia Erica Yehezkiel

    Technology development is very rapid, so all fields are required to develop technology to increase the effectiveness and efficiency of work. One of the focuses is related to image processing technology. We can get many benefits by implementing this system, so various fields have implemented image processing systems, such as security, health, and education. One of the current obstacles is in the area of safety, namely in the field of searching for people, which is still done manually. Often search teams find it challenging to find people because of the significant search area, low light conditions, and complex search fields. Therefore, we need a tool capable of detecting humans to assist in finding people. Therefore, to detect human objects, the authors try to research human object detection using a simple device for the human object detection system. The authors use the You only look once (YOLO) method with the YoloV4-Tiny type, where this algorithm has high detection speed and accuracy. Using the YOLOV4-Tiny simulation method for human object recognition, a detection rate of 100% is obtained with an FPS value of 5.


  • Open Access

    Fuzzy Mamdani performance water chiller control optimization using fuzzy adaptive neuro fuzzy inference system assisted

    by Galang Persada Nurani Hakim, Rachmat Muwardi, Mirna Yunita, Diah Septiyana

    Fuzzy Mamdani knows as one of the modern control systems. It was known to have a better performance result when compared to conventional methods. However, because the input of this modern control system sometimes is based on human experience, therefore its performance is sometimes below the conventional one. We propose using the adaptive neuro fuzzy inference system assisted (ANFIS) approach to optimize the fuzzy Mamdani membership function input to overcome this problem. We have tested our hypotheses in water chiller applications based on microcontrollers. Even though it is still behind conventional methods to cool 200 ml water, which is 6 minutes, using fuzzy ANFIS methods, we manage to improve the speed performance in cooling water from 20 minutes to 17 minutes, which is from room temperature to just 24 oC.


  • Open Access

    wisata Pengembangan Desa Wisata di Kelurahan Kemumu Kecamatan Arma Jaya Kabupaten Bengkulu Utara

    by Mirna Yunita, Warsa Sugandi, Edwar Edwar, Yuneva Syukur, Fevi Wira Citra, Nurmintan Silaban

    Against the background, the potential for tourism has not been managed into a single tourism package that can contribute to Local Original Revenue if it is managed properly. The purpose of this study is to formulate alternative strategies for developing tourism villages in Kemumu Village, Arma Jaya District, North Bengkulu by making physical, cultural and artificial potential into an attractive tour package. This type of research is a qualitative descriptive type conducted at the Office of Tourism and Culture of North Bengkulu Regency, Muara Bangkahulu District, Society and tourists. Determination of the informant is done purposively. Data collection is done through observation, interviews, and documentation. Data management is done through SWOT analysis. The results of the study can be formulated by six development strategies that the government must undertake: 1) improving access roads for connecting with other regions; 2) building supporting facilities around attractions (such as toilets, places of worship, restaurants, temporary dwellings, villas, parking lots and culture areas); 3) provide a place for cultural arts performances; 4) cooperating with farmers regarding permits to use terraced rice terraces or terraces for tourism activities (such as: tourists may participate in farming activities); 5) cooperating with private parties; 6) provide training to the community.

  • Research and Design of Fast Special Human Face Recognition System

    by Rachmat Muwardi, Huangyao Qin, Hongmin Gao, Harun Usman Ghifarsyam, Muhammad Hafizd Ibnu Hajar, Mirna Yunita

    During the use of face recognition systems appears the problems such as the system can not find a face and has not enough information of the image, changing illumination, occlusion, face shape changes and so on. In order to solve the current problems and improve the ability of face recognition and detection, new methods for data acquisition and some tools are adopted, and the application development of improved face recognition system is proposed based on ARM architecture design, software development of face recognition system is taken in ARM embedded platform. The solve project for the acquisition of three-dimensional face data, biometric recognition, computer vision, computer graphics is researched and designed. The image edge contour detection method is used for face image features positioning. The adaptive feature extraction method of image entropy is used for facial feature extraction, the face recognition algorithm is optimized, and the face image output module is constructed. The simulation results show that the face recognition system designed in this paper has good robustness performance, and it can improve the accurate probability of face recognition.