Pocholo James M. Loresco
30 Publications
Scopus ID: 105034159397
TENCON 2025 - 2025 IEEE Region 10 Conference (TENCON), (2026), pp. 1215-1219
Conference Paper | Published: February 18, 2026
Abstract
Filipino Sign Language (FSL) is an invaluable tool for communication within the deaf and mute communities, yet there is a shortage of proficient special education teachers and accessible learning materials. Current research on FSL recognition is limited to basic detection, often invasive, and lacks comprehensive systems that provide feedback to users. Additionally, FSL incorporates distinctive static and dynamic gestures, including contractions, which set it apart from other sign languages. This study presents the development of a machine vision-based FSL tutor that leverages the MediaPipe framework-specifically, MediaPipe Hands for static gesture recognition and MediaPipe Holistic for full-body dynamic gesture tracking. LSTM networks were used to classify dynamic gestures based on sequential landmark data to capture temporal dependencies in sign execution. The system supports a desktop application platform enabling learners to engage in interactive modules with real-time feedback through visual prompts and audio cues. It utilizes 42 static hand feature landmarks and over 1,662 key points derived from hand, pose, and facial data to ensure accurate recognition and feedback. A total of 50 essential FSL gestures-aligned with the kindergarten curriculum-were modeled, covering alphabet knowledge, vocabulary development, self-introduction, and polite expressions. Performance evaluation using computer vision metrics demonstrated high recognition accuracy for both gesture types. In addition, the System Usability Scale (SUS) and statistical comparisons with traditional instruction methods confirmed the platform's effectiveness and user acceptability. The results validate the system as a comprehensive and accessible solution for FSL education, particularly suited for early learners and self-guided instruction.
Scopus ID: 105000426847
TENCON 2024 - 2024 IEEE Region 10 Conference (TENCON), (2024), pp. 842-845
Conference Paper | Published: January 1, 2024
Abstract
Visual impairment continues to be a global health concern. People with visual impairment experience difficulty moving around indoors, especially in unfamiliar spaces. While existing assistive technologies like smart canes offer point-to-point navigation or rely on infrastructure like RFID tags or beacons, they lack the ability to provide comprehensive indoor navigation with obstacle detection and avoidance. This paper presents a novel indoor navigation system for visually impaired individuals using deep learning and audio guidance. The system utilizes 3D-printed glasses equipped with a Raspberry Pi v2 camera, audio user interface and a processing unit comprising a Raspberry Pi 4B and Google Coral Edge tensor processing unit (TPU). As validated in a controlled indoor environment, the deep learning models for localization, navigation, obstacle detection, and obstacle avoidance achieve high results in terms of accuracy, precision recall, and F1-score. Based on user tests using the System Usability Scale, this wearable assistive device appears to offer a promising solution for promoting independent navigation and spatial awareness among visually impaired individuals.
Scopus ID: 105000234438
TENCON 2024 - 2024 IEEE Region 10 Conference (TENCON), (2024), pp. 1286-1289
Conference Paper | Published: January 1, 2024
Abstract
Outcome-Based Education (OBE) emphasizes the importance of defining and assessing specific learning outcomes. Effective monitoring of these outcomes is crucial for ensuring student success and program effectiveness. Previous research has explored various approaches to enhance program outcome monitoring, however, have not fully addressed the need for individualized and comprehensive progress tracking that goes beyond binary pass or fail measurements. This paper presents a novel approach to enhance program outcome monitoring through the development of individualized dashboards and the application of an adaptive neuro-fuzzy logic (ANFIS) framework. Data were derived from CSV reports of students in a learning management system and Canvas New Analytics from a sample class in the pilot study. The ANFIS framework is based on formative and summative assessments, total and maximum page views and participation, and average weekly page views and participation. The ANFIS model and dashboard results demonstrate its effectiveness in providing students and educators with a deeper understanding of student progress in terms of program outcomes, enabling targeted interventions and personalized learning experiences. This comprehensive approach empowers educators with the tools and insights needed to optimize educational practices and ensure that all students achieve the desired learning outcomes.
Scopus ID: 85199862569
2023 IEEE 15th International Conference on Humanoid, Nanotechnology, Information Technology, Communication and Control, Environment, and Management (HNICEM), (2023), pp. 1-6
Conference Paper | Published: January 1, 2023
Abstract
The transcription of music is essential since it preserves the originality of the music and the compositional technique used by the composers. In this manner, native music can be reproduced, and documents of certain tunes can be passed on to succeeding generations without hearing the original music. A very limited amount of research has been conducted on the application of automation in music transcription using deep learning, particularly in native music instruments. This research is an effort to preserve and conserve Filipino culture in the context of native music and musical instruments particularly the Diwdiw-as, a native Filipino bamboo flute. Using signal processing and deep neural networks, the proposed study aims to automate music transcription. The system is capable of classifying pitches based on the fundamental frequency and is also capable of classifying notes based on their duration. Diwdiw-as pitch can be transcribable to the music sheet using the following pitches: CS, DS, ES, FS, GS, AS, BS, C6, while the note values are whole, eighth, quarter, half, dotted eighth, dotted half, dotted quarter. A web-based application has been developed to assist in the automatic music transcription (AMT) of Diwdiw-as. A pdf file of the transcribed music sheet can then be downloaded. According to the confusion matrices, the system's accuracy is high in terms of the transcription of pitches and notes in the music sheet.
Scopus ID: 85199876371
2023 IEEE 15th International Conference on Humanoid, Nanotechnology, Information Technology, Communication and Control, Environment, and Management (HNICEM), (2023), pp. 1-6
Laurence Keith P. Alquiza, King Harold A. Recto, ... Jesús Villalobos
Conference Paper | Published: January 1, 2023
Abstract
The Republic of the Philippines is confronted with rebuilding its energy landscape, which now depends heavily on imported fossil fuels for a substantial supply. The Department of Energy (DOE) has established lofty objectives to enhance the nation's renewable energy (RE) capability; nevertheless, these objectives are still to be achieved. This research supports DOE's goals by studying other possible renewable energy sources. In particular, the primary aim of this research is to examine the viability of Offshore Wind Converters (OWCs) and Wave Energy Converters (WECs) as viable sustainable energy options for the Philippines. Ocean wave converters (OWCs) provide inherent benefits in terms of dependability and have widespread societal acceptance. Conversely, wave energy converters (WECs) harness the vast energy potential contained within ocean waves. A comparative evaluation was undertaken to analyze the differences between these two potential renewable energy sources. The assessment concludes that OWCs possess a minor advantage over WECs regarding their economic viability and higher societal acceptability. It recommended that the government adopts a diversified energy portfolio, which may include the incorporation of WECs to effectively navigate the changing dynamics of the energy sector, enhance sustainability, and ensure the long-term security of the nation's energy supply.
Scopus ID: 85199891103
2023 IEEE 15th International Conference on Humanoid, Nanotechnology, Information Technology, Communication and Control, Environment, and Management (HNICEM), (2023), pp. 1-6
Conference Paper | Published: January 1, 2023
Abstract
The energy efficiency of cell sites is of increasing importance due to the proliferation of wireless communication and the energy-intensive nature of their operation. Telecommunication macro cell sites provide coverage over large areas and contain various equipment, such as base stations, antennas, radio frequency (RF) components, transmission systems, and power infrastructure. These sites require active cooling elements controlled by microprocessors, making their energy efficiency an essential issue to address. The Power Usage Effectiveness (PUE) metric is a widely accepted assessment for energy efficiency within data centers. It has been applied in numerous studies, but there is a significant gap in research regarding its application in telecommunication macrocell sites. Due to the potential applications of PUE beyond data centers, this study proposes to extend this metric to telecommunications macrocell sites. In this study, the PUE-based operational efficiencies of telecommunication macro cell sites in the Philippines are determined using neurofuzzy inference method based on the collected data, which includes total grid consumption, genset AC consumption, facility power consumption, and DC power consumption. The study utilized data gathered from telecommunication macrocell sites situated in various regions of the Philippines, namely North Luzon, South Luzon, Visayas, and Mindanao. The study considered four different categories of telecom macro cell sites, namely Class A, B, Cl, and C2. A Sugeno type adaptive neuro-fuzzy inference (ANFIS) model is constructed based on exhaustive search using the two most influential input variables. A Root Mean Square Error analysis of the proposed model shows promising results.
Scopus ID: 85199888295
2023 IEEE 15th International Conference on Humanoid, Nanotechnology, Information Technology, Communication and Control, Environment, and Management (HNICEM), (2023), pp. 1-6
Conference Paper | Published: January 1, 2023
Abstract
The coffee rust disease threatens coffee production in the Philippines with widespread defoliation and reduced yield. Identifying rust infection and its severity is critical for implementing effective mitigation strategies. As an alternative to recent methods that rely on deep learning approaches, our vision-based approach utilizes Artificial Neural Networks, K-Nearest Neighbors, and Thresholding methods to identify rust infection on coffee leaves and estimate severity, providing a computationally lightweight alternative for agricultural disease management. Twenty-four (24) color and texture features of a collected dataset of coffee leaf images were extracted as inputs for an ANN classifier. The percentage of damage on coffee leaves was determined by comparing the damaged pixels to the total area of the leaf using KNN and thresholding segmentation techniques. Through the use of confusion matrix and RMSE, the decision support system has demonstrated promising results in identifying coffee leaf health and estimating severity of coffee rust infection.
Scopus ID: 85145856105
Journal of Advanced Computational Intelligence and Intelligent Informatics, (2022), Vol. 26, No. 6, pp. 905-913
Ryan Rhay P. Vicerra, Argel A. Bandala, ... Alvin Culaba
Journal Article | Published: November 1, 2022
Abstract
Due to the advent of the COVID-19 pandemic, the Philippine government encouraged enterprises and businesses to utilize flexible work arrangements such as work-from-home (WFH) or telecommuting setup. Nowadays, the key components necessary for a telecommuting include a WiFi-enabled IT equipment, secured work environment, and reliable internet connection, while research shows that type of work and computer literacy are also key factors for telework implementation. Multiple studies in relation to telework have already been conducted but some studies were deemed inconclusive and need further analysis. Therefore, in this study, a Mamdani fuzzy-based model was developed for telework capability assessment for Philippine government employees based on four significant factors namely: internet speed, IT equipment availability, computer literacy, and type of work, which are expressed in linguistic representations. The proposed fuzzy system can provide a feedback telework capability score based on the four input parameters which may also be characterized with the potential telecommuting cost requirement.
Scopus ID: 85141956973
DOI: 10.1063/5.0072420
AIP Conference Proceedings, (2022), Vol. 2433, pp. 030004
Conference Paper | Published: October 26, 2022
Abstract
Engineering students face challenges of on-time successful degree completion. Universities incorporate academic advising as a solution to these challenges. Decision support systems enhance the effectivity of academic advising. Combined with machine learning, it can predict future student performance providing useful information. Compared to common ‘black box’ models, linguistic rules provide better interpretation and insight discovery. However, existing models often use positive predictors of academic excellence, with limited consideration on factors of negative effect. This work, therefore, generates linguistic rules for academic advising based on three predictors using rough set theory (RST) and then compared with artificial neural network (ANN) for benchmarking. Forty-eight samples of mechanical engineering students taking up machine design were considered. RST attained accuracy of 72.92% while ANN attained 66.66%. The model generated 13 linguistic rules, having reflected unrealized insights. The findings from this study may be utilized by academic advisers for pattern recognition, in identifying ‘at-risk’ students.
Scopus ID: 85142100199
2022 IEEE 10th Region 10 Humanitarian Technology Conference (R10-HTC), (2022)
Conference Paper | Published: January 1, 2022
Abstract
The incidence of falls is more commonly experienced by the ageing population globally, due to their increasing frailty. Fall detection systems are being developed combined with machine learning approaches that include those wearable devices, ambience-based systems, and vision-based systems. Wearable systems are omnipresent but intrusive while ambience-based systems are temperature-dependent that require a dedicated GPU. Challenges on vision-based fall detection systems include large computational cost, use of specialized camera, and system integration into smartphones. This research aims to propose a vision-based fall detection system ported on a smartphone application, that utilizes deep learning trained and tested from multiple RGB camera setups. In an event of a fall, the system is designed to provide an on-premises auditory alarm, an IoT notification, and a real-time video feed via the smartphone application. Using the pretrained MobileNetv2 CNN-based MediaPipe Pose and Random Forest Classifier for fall detection, experimental results show high-performance evaluation metrics.