MFMuhammad Fachrurrozi

Muhammad Fachrurrozi

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Muhammad Fachrurrozi has a total of 11 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Enhancing Real-Time Face Detection Performance Through YOLOv11 and Slicing-Aided Hyper Inference

    by Muhammad Fachrurrozi, Muhammad Naufal Rachmatullah, Akhiar Wista Arum, Fiber Monado

    Real-time face detection in crowded scenes remains challenging due to small-scale facial regions, heavy occlusion, and complex illumination, which often degrade detection accuracy and computational efficiency. This study presents an enhanced detection framework that integrates Slicing-Aided Hyper Inference (SAHI) with the YOLOv11 architecture to improve small-face recognition under diverse visual conditions. While YOLOv11 provides a high-speed single-stage detection backbone, it tends to lose fine spatial information through downsampling, limiting its sensitivity to tiny faces. SAHI addresses this limitation by partitioning high-resolution images into overlapping slices, enabling localized inference that preserves structural detail and strengthens feature representation for small targets. The proposed YOLOv11-SAHI system was trained and evaluated on the WIDER Face dataset across Easy, Medium, and Hard difficulty levels. Experimental results demonstrate that the integrated framework achieves Average Precision (AP) scores of 96.33%, 95.87%, and 90.81% for the respective subsets-outperforming YOLOv7, YOLOv5, and other lightweight detectors, and closely approaching RetinaFace accuracy. Detailed error analysis reveals that the combined model substantially enhances small-face detection in dense crowds but remains sensitive to severe occlusion, motion blur, and extreme pose variations. Overall, YOLOv11 coupled with SAHI offers a robust and computationally efficient solution for real-time face detection in complex environments, establishing a foundation for future work on pose-invariant feature learning and adaptive slicing optimization.

  • Open Access

    English – Indonesian Phrase Translation Using Recurrent Neural Network and ADJ Technique

    by Wenty Octoviani, Muhammad Fachrurrozi, Novi Yusliani, Mukhlis Febriady, Afriyan Firdaus

    Recurrent Neural Network (RNN) and annotated disjunction are discussed to develop English – Indonesian phrase based. Phrase – based English – Indonesian machine translation becomes important because there are differences between two languages and it can be a sub system for English – Indonesian machine translation. Automatic translation can be done using RNN and ADJ Technique. The main process in this research are preprocessing, determination of phrase type, and translation the phrase. In the preprocessing has process such as case folding, tokenizing, pos tagging, and stemming. In the determination of phrase type, system counts weight of input phrase using RNN. The translation phrase is entered to target language using ADJ Technique. The software experiment is tested by using 70 English phrases. The result shows accuracy of RNN and ADJ Technique for English – Indonesian phrase translation is 88.57 %.


  • Open Access

    Faster R-CNN with Inception V2 for Fingertip Detection in Homogenous Background Image

    by Derry Alamsyah, Muhammad Fachrurrozi

    Fingertip Detection is a versatile research field in computer vision, since it has multiple purpose such as natural user interface, robotic, 3D simulation etc. It is a challenging research field in computer vision. There are some hand segmentation methods, pre-processing phase, was conducted to provide area of fingertip detection in image. However, in this research, fingertip detection can be done by directly find the fingertip itself. This approach cut off preprocessing phase by using Faster R-CNN method and inception V2 architechture directly to find the fingertip in image. With a homogenous background as a simple input image, this approach showed a good accuracy in its performance. It has 90% and 91% accuracy in way to detect fingertip for both male and female hand datasets. More over, exchanging male and female model toward to male and female dataset gave 94% and 92% accuracy that showed the different pattern between both.


  • Open Access

    Real Time Monitoring System of Pollution Waste on Musi River Using Support Vector Machine (SVM) Method

    by Muhammad Fachrurrozi, Erwin Saparudin, Erwin

    Real-time Monitoring and early detection system which measures the quality standard of waste in Musi River, Palembang, Indonesia is a system for determining air and water pollution level. This system was designed in order to create an integrated monitoring system and provide real time information that can be read. It is designed to measure acidity and water turbidity polluted by industrial waste, as well as to show and provide conditional data integrated in one system. This system consists of inputting and processing the data, and giving output based on processed data. Turbidity, substances, and pH sensor is used as a detector that produce analog electrical direct current voltage (DC). Early detection system works by determining the value of the ammonia threshold, acidity, and turbidity level of water in Musi River. The results is then presented based on the level group pollution by the Support Vector Machine classification method.


  • Open Access

    Real Time Monitoring System of Pollution Waste on Musi River Using Support Vector Machine (SVM) Method

    by Muhammad Fachrurrozi, Saparudin Saparudin, Erwin Erwin

    Real-time Monitoring and early detection system which measures the quality standard of waste in Musi River, Palembang, Indonesia is a system for determining air and water pollution level. This system was designed in order to create an integrated monitoring system and provide real time information that can be read. It is designed to measure acidity and water turbidity polluted by industrial waste, as well as to show and provide conditional data integrated in one system. This system consists of inputting and processing the data, and giving output based on processed data. Turbidity, substances, and pH sensor is used as a detector that produce analog electrical direct current voltage (DC). Early detection system works by determining the value of the ammonia threshold, acidity, and turbidity level of water in Musi River. The results is then presented based on the level group pollution by the Support Vector Machine classification method.