ASEAN Journal of Scientific and Technological Reports

Application of YOLOv8 Improved Model Incorporating Attentional Mechanism in Weed Identification

Zhanwei Feng — 2026-03-24

Accurate identification of weeds plays a pivotal role in precision agriculture, as it supports targeted herbicide application and enhances crop productivity. This paper presents an improved YOLOv8-based object detection framework that integrates a global attention mechanism to strengthen weed recognition performance. Although YOLOv8 demonstrates strong baseline accuracy and speed, it lacks sufficient global contextual modeling, which constrains its ability to capture fine-grained features in visually complex or dense field scenarios. To overcome this limitation, we incorporate a Global Attention Module (GAM) after each C2f block in the backbone, enabling the network to capture long-range dependencies and highlight semantically relevant regions. Experiments on a custom weed dataset indicate that the attention-enhanced model considerably surpasses the baseline YOLOv8 in precision, recall, and mean Average Precision (mAP), especially under challenging conditions with small-scale or overlapping targets. Furthermore, the proposed method attains a favorable trade-off between detection accuracy and computational efficiency, making it suitable for real-time smart farming applications.

Improving Drying Efficiency for Red Chili Peppers Using Ventilated Solar Technology

Tanawat Srirugsa — 2026-03-24

This study compares three drying methods for Jinda red chili peppers: open sun drying (OSD), conventional solar drying (SD), and ventilated solar drying (VSD). Fresh chilies with an initial moisture content of 80% were dried to achieve the industry standard of ≤13.5% final moisture content. Temperature monitoring showed that SD reached the highest peak temperature (75°C), followed by VSD (56°C) and OSD (52°C). Significant differences in drying time were observed: VSD reached appropriate moisture levels in under 17 hours, SD in 34 hours, and OSD in 45 hours. Compared with the first day of SD and OSD, which had drying rates of 49.5% and 55.5%, respectively, VSD showed the highest drying rate, reducing the moisture content to 34.6%. By the second day, VSD had a moisture content of 12.5%, while SD and OSD retained 33.7% and 37.3%, respectively. The enhanced efficiency of VSD is attributed to the combined effects of elevated temperature and mechanical ventilation, which accelerated moisture removal despite the lower chamber temperature than in SD. While solar dryers require greater initial investment than traditional methods, they offer significant advantages in processing time, product quality, contamination prevention, and labor requirements, potentially yielding faster returns on investment. These findings suggest that VSD is optimal for commercial chili drying operations in Thailand's agricultural sector.

Technology Transfer Assessment of Amino Acid Block Fertilizer Production from Salted Egg White Waste for Coconut Farming in Paktrae, Songkhla Province

Saowanee Lekbangpong — 2026-03-24

This study aimed to investigate farmers’ knowledge of Good Manufacturing Practice (GMP)-compliant production of salted egg products, evaluate the technology transfer process, and assess user satisfaction with the application of amino acid block fertilizer derived from salted egg white waste for coconut cultivation. The study sample comprised 25 salted egg producers from Paktrae Subdistrict, Ranot District, Songkhla Province, Thailand. Data were collected using structured questionnaires and analyzed using descriptive statistics, including percentages, mean, and standard deviation, as well as a paired t-test. The results indicated that before the technology transfer, farmers exhibited a low level of knowledge, with a mean score of 5. After the technology transfer, knowledge increased significantly, with a mean score of 14, and the difference was statistically significant at the 0.01 level. The technology transfer process for amino acid block production from salted egg white waste was implemented through a participatory approach, including GMP training, technology introduction, hands-on practice, participatory discussions on product labeling, and the development of user manuals and infographic-based labels. Overall satisfaction with the technology transfer was at the highest level, with a mean score of 4.30. The highest satisfaction ratings were associated with lecturers (4.66), staff services (4.65), and knowledge and understanding gained from the training (4.48), followed by knowledge application (4.32) and training location (4.15). These findings suggest that participatory technology transfer can effectively enhance farmers’ knowledge and satisfaction. Continuous monitoring and sustained technology support are recommended to ensure effective adoption and long-term utilization of the transferred technology among target communities.

Orbital Period Change of the Binary Star System V829 Aql

Supat Sairattanain — 2026-03-24

A variable star, V829 Aql, is considered one of the binary stars providing invaluable insights into the process of stellar evolution. The purpose of this research study is to calculate the orbital period change of the binary star system V829 Aql using a telescope at the Regional Observatory for the Public, Songkhla, National Astronomical Research Institute of Thailand (Public Organization). Images in the B and V wavelength range will be taken on the night of July 1-2, 2021, from 9:00 p.m. to 5:00 a.m., and subsequently examined. It was discovered that the brightness value pattern adhered to an equation (relationship): the orbital period equals the lowest brightness when it does. The orbital period change is growing, as indicated by the equation O-C = (4.82878 Í 10^-13) Epoch² - (1.5221 Í 10^-7) Epoch – 0.01052, which gives a rate of change in orbital period of 1.05 x 10^-4 seconds/year. This means that stars continue to separate, which explains the evolution of the TRO theory.

AI-Based Detection of Leaf Diseases in Durian (Durio zibethinus) Using Convolutional Neural Networks: Model Development and Performance Evaluation

Supreeya Thepparak — 2026-03-24

Early detection of durian leaf diseases using artificial intelligence (AI) plays an important role in supporting effective plant disease management within digital agriculture systems. This study aimed to develop an AI-based system for the preliminary detection and classification of common durian leaf diseases using image analysis. Diseased leaf samples were collected from durian orchards in Pa Phayom District, Phatthalung Province, and Chang Klang District, Nakhon Si Thammarat Province, southern Thailand. Associated microorganisms were isolated using the tissue transplanting technique, and morphological characteristics indicated the presence of Colletotrichum spp. in leaf spot samples and Fusarium spp. in leaf blight samples. In contrast, the algal leaf spot pathogen could not be cultured on standard fungal media. A Convolutional Neural Network (CNN) was employed to classify three disease categories: leaf spot, leaf blight, and algal leaf spot. The image dataset was divided into training, validation, and testing sets, and the model was trained and evaluated accordingly. The results demonstrated that the proposed AI model effectively classified durian leaf disease images, particularly leaf spot and leaf blight. However, this study used a relatively limited dataset, and the proposed system should therefore be considered a preliminary disease-detection tool. The developed AI-based system shows potential for deployment on mobile devices or online platforms to support early disease recognition and decision-making by farmers, thereby improving disease management strategies in durian cultivation.

Work-related Musculoskeletal Disorders among Rubber Tappers in Phatthalung Province, Thailand

Sutee Inraksa — 2026-03-24

This cross-sectional study presents the prevalence of Work-related Musculoskeletal Disorders (MSDs) among rubber tappers in Pa Phayom District, Phatthalung Province, Thailand. The results were assessed using descriptive statistics and binary logistic regression at the 95% confidence interval, with consideration of variables influencing MSDs. The results showed that most participants (51.90%) were female, with a sample mean age of 43.95 ± 10.77 years, an average work experience of 13.68 ± 8.87 years, and a working time per day of 3.75 ± 1.52 hr. The physical work characteristics of the sample group revealed that the rubber plantations were on flat land (87.01%), more than 500 rubber trees were tapped per day (62.99%), the tapping height was below waist level (61.04%), and the rubber trees were older than 10 years (73.38%). The results for the severity of musculoskeletal disorders at the unbearable level showed that the lower back had the highest incidence (9.1%), followed by the upper back (2.6%). At the severe level, the lower back (26.6%) had the highest incidence, followed by the shoulder (7.8%) and the upper back (7.1%). When considering overall pain levels across different body parts, the hand/wrist was the most common (87.7%), followed by the neck (83.1%) and the lower back (82.5%). Factors associated with MSD symptoms among rubber farmers include the age of the trees and exercise. The reported acute MSD in their lower backs throughout the one-month study duration. Therefore, ergonomic knowledge should be promoted to help rubber farmers improve their working posture and reduce injuries.

SUMO-Based Optimization of Intersection Performance under Heterogeneous Traffic

Md. Emtiaz Kabir — 2026-03-24

Traffic congestion is one of the most critical challenges in metropolitan cities, where intersections often experience long queues, high delays, and increased emissions due to mixed traffic and reliance on manual police control. This study examined how geometric improvements and signal timing optimization can improve the performance of a busy four-arm intersection in Uttara, Dhaka. Traffic data were collected through field surveys during peak hours and used to build a detailed microsimulation model in SUMO, calibrated with local driving behavior. Three scenarios were tested: the existing manual police control, channelization of turning lanes with current timings, and channelization with optimized fixed-time signals using Webster’s method. Results showed that under current conditions, delays and queues were extremely high, corresponding to the lowest service level. Channelization alone reduced delays by up to 40% and improved overall flow, while the combination of channelization and optimized signals achieved the best outcomes, cutting delays to less than 40 seconds per vehicle, reducing queues by more than 70%, and improving the level of service to stable and acceptable conditions. Environmental analysis confirmed that the optimized scenario lowered fuel consumption and carbon emissions by 25–35% compared to current operations. These findings demonstrate that integrating physical improvements with evidence-based signal control can substantially enhance traffic efficiency, reduce energy use, and improve air quality in rapidly growing cities. The study provides practical insights for policymakers and planners seeking sustainable, low-cost solutions to congestion in Dhaka and other megacities facing similar challenges.

Production of Fermented Vinegar Drink From Hang Rice With Pineapple Peel Syrup

Nisa Romsomsa — 2026-03-24

This study aimed to develop fermented vinegar beverages produced from Hang rice combined with pineapple peel syrup through traditional fermentation. The fermentation process was carried out under ambient conditions for 30 days using a traditional method to monitor changes in acetic acid bacteria and yeast populations, as well as acetic acid production. Consumer preference for the final product was also assessed. The microbial populations of both acetic acid bacteria and yeast showed a sharp increase during the first 14 days, reaching their peaks at 7.92 ± 0.24 log CFU/mL and 7.06 ± 0.28 log CFU/mL, respectively, after 30 days. The final vinegar exhibited 4.44 ± 0.31% acetic acid, a pH of 3.6, and showed no detectable alcohol. Three formulations were prepared by varying the ratio of Hang rice vinegar to pineapple peel syrup—Formula 1 (60:40), Formula 2 (50:50), and Formula 3 (40:60). Among them, Formula 3 demonstrated the highest DPPH radical scavenging activity (503.70 ± 24.42 µg Trolox/mL) and the greatest total phenolic content (97.24 ± 4.36 µgGAE/mL). Sensory analysis revealed comparable acceptability scores across all samples, ranging from “like slightly” to “like moderately”. These findings suggest that Hang rice vinegar supplemented with pineapple peel syrup represents a novel beverage with significant in vitro antioxidant capacity, adding value to germinated brown rice and promoting the utilization of pineapple peel waste.

Comparison of Color Space Performance for Colorimetric Detection of Heavy Metals in Drinking Water Using Image Processing and Convolutional Neural Networks

Atthapol Phongsuwan — 2026-03-24

Heavy metal contamination in drinking water, including arsenic, lead, cadmium, iron, copper, and manganese, is an environmental issue that adversely affects human health and ecosystems. Although standard laboratory methods provide high accuracy, they are limited by cost, time, and operational complexity. This study aimed to develop and compare the performance of eight color spaces (RGB, RGBA, Grayscale, GrayscaleA, CIE LAB, CIE LABA, HSV, HSVA) for classifying heavy metal concentration levels from microplate images using image processing techniques combined with a Convolutional Neural Network (CNN), and to select the optimal color space using the Weighted Sum Model. The dataset comprised images of six types of heavy metals, with 368,000 images after preprocessing and data augmentation. The model was trained using K-Fold Cross Validation (K=5). Experimental results showed that the HSVA color space achieved the best performance (Accuracy 99.65%, Loss 0.01382, Training Time 1,606 seconds). When tested on a separate set of 920 images, the model maintained an accuracy of 99.0%, indicating stability and strong practical applicability. The findings confirm that selecting an appropriate color space in conjunction with a CNN significantly improves the accuracy of heavy metal analysis on test datasets and has strong potential for further development into a portable, low-cost analytical tool for field applications.

Effects of Cassava Tails and Stalk and Superabsorbent Polymer on Soil Properties, Growth, and Yield of Cassava in Sandy Loam and Clay Loam Soils

Prinda Kongyinghan — 2026-03-24

The objective of this study was to evaluate the effects of cassava tails and stalk (CTS) and superabsorbent polymer (SAP) on soil properties, as well as on cassava growth and yield in sandy loam and clay loam soils. The research employed a randomized complete block design with three treatments, including unamended soil (control, T1) , 8,000 kg/rai of fresh weight CTS (T2), and 0.25% (w/w) of SAP (T3) with four replications. In sandy loam soil, the SAP application showed a significant decrease in soil bulk density (BD) and increase in soil particle density (PD) and soil porosity (PO). Both soil amendment treatments increased organic matter (OM) (0.58%) compared with the control (0.51%). The SAP application showed the highest content of P and K (9 and 69 ppm, respectively), while CTS showed the highest content of Ca (411 ppm). In clay loam soil , the SAP application resulted in the lowest BD (1.36 g/cm³), while the CTS application resulted in the highest soil PO (50.11%). Both treatments increased OM (~1.00%) and the content of Ca and Mg (~2,000 ppm and ~220 ppm, respectively) compared with the control (0.73%, 1,559 ppm, and 182 ppm, respectively). The SAP application showed a better leaf water potential and stomatal conductance, which led to the highest plant height, leaf area, and stem diameter of cassava. In addition, the SAP application resulted in the highest cassava yield (7.83 and 6.20 tons/rai) in both soils and starch content (21.44%) in sandy loam soil. Therefore, both SAP and CTS applications improved soil properties and promoted cassava growth, resulting in higher yields; however, their effectiveness was strongly dependent on soil texture.

The Influence of Managerial and Operational Characteristics on Food Waste Reduction Practices Among the Restaurants Along the Andaman Coast of Thailand

Mumtas Meraman — 2026-03-25

The food service sector plays a substantial role in the food waste creation. Restaurants in Thailand, driven by tourist demand, substantially contribute to food waste. This study analyzes the managerial and operational attributes influencing restaurants' practices for reducing food waste in Thailand's Andaman coastal region. This quantitative research uses the Waste Hierarchy framework to investigate restaurants' food waste practices. Independent variables are divided into two categories: 1) Manager Characteristics, and 2) Restaurant Characteristics. In total, 353 questionnaires were distributed using proportional allocation. The data were collected via simple random sampling and analyzed using logistic regression. The results indicate that restaurants demonstrated a high level of compliance with prevention practices; however, they showed substantially lower adoption of reuse and recycling practices. The logistic regression models revealed that personal factors (education, age, interest in food waste reduction, and restaurant experience), along with the number of meals served at the restaurant, significantly influence food waste practices. The multiple logistic regression model showed a good fit (model chi-square = 72.78, df = 6, p < .0001) and explained 25.2% of the variance. This study offers valuable insights and practical implications by highlighting that manager characteristics, rather than operational attributes, are important in shaping food waste reduction. This insightful information can be used to design and implement policies on human capacity development related to food waste reduction.

Comparative Analysis of Rubberwood, Plywood, and Wood-Plastic Composites for Safe Elderly Toy Applications

Pruttipong Pantamanatsopa — 2026-03-25

The objective of this research was to produce toy products for the elderly to develop the potential in various areas of the elderly. Different types of wood materials, namely rubberwood sawn timber, plywood, and wood-plastic composites, were evaluated for their performance based on mechanical, physical, and chemical properties. Samples were compressed using a compression molding machine. The experimental results were analyzed by Analysis of Variance (ANOVA). The findings showed that all mechanical properties of the elderly toy products manufactured from rubberwood (RW) had higher strength values than those of plywood (PW) and wood-plastic composites (WPCs). The lowest water absorption and thickness swelling values of 4.36% and 2.01%, respectively, were observed in elderlytoy products manufactured from WPCs. In contrast, the minimum Ra value of 1.45 µm was observed. Additionally, the results showed that the three types of materials contain heavy metal content that does not exceed the specified standards, indicating that this product is safe for the elderly. Finally, the elderly toy products made from WPCs showed smooth surfaces, greater agglomeration, and fewer pores and voids, due to the strong interfacial adhesion between the plastic matrix and the rubberwood fiber reinforcement.

The Preparation and Characterisation of a Bioplastic Film from Acrylated Epoxidized Soybean Oil (AESBO)

Apatsara Uchupap — 2026-03-25

This study successfully demonstrates the synthesis of epoxidized soybean oil (ESBO) via reaction with formic acid and hydrogen peroxide, followed by acrylation with varying ratios of acrylic acid. The characterization techniques, including FT-IR and ¹H-NMR, confirmed the formation of key functional groups such as oxirane and hydroxyl, as well as the reduction of carbon–carbon double bonds, indicating successful epoxidation and acrylation. The appearance of new peaks in the 1H-NMR spectra further supports the formation of the epoxide and acrylate functionalities. FT-IR analysis revealed clear spectral changes correlating with these transformations. The kinetics study using 5 wt% photoinitiator demonstrated that the acrylated epoxidized soybean oil (AESBO) film could be rapidly stabilized within 60 seconds under UV curing. These findings highlight the potential of AESBO as a sustainable precursor for bioplastic films, offering promising chemical stability. Further optimization may enhance its suitability for various biodegradable polymer applications.

Evaluation of the Nutritional, Mineral, and Estimated Metabolizable Energy Values of Agricultural Biomass Raw Materials Used for the Development of Concentrated Fermented Feed for Beef Cattle in Mukdahan Province, Northeast Thailand

Nathaporn Jirawattanasomkul — 2026-03-25

This study evaluated the nutritional, mineral, and estimated metabolizable energy (EME) profiles of eight locally available agricultural biomass raw materials to support the development of a cost-effective concentrated fermented feed (CFF) for beef cattle in Mukdahan Province, Northeast Thailand. Finely ground maize kernels, soybean meal, palm kernel meal, rice bran, dried cassava chips, cassava pulp, washed cassava peels, and fresh cassava starch residue were analyzed for proximate composition, EME, and essential minerals (Ca, Mg, P, Fe, Zn, Na, and K), with all values reported on an as-fed basis. A CFF formulation (1,000 kg batch) was developed from these raw materials and compared with five commercial beef cattle feeds. Soybean meal, palm kernel meal, and rice bran exhibited the highest protein contents, while rice bran showed the greatest calculated EME value (4097 kcal/kg of sample). Cassava-based by-products were rich in carbohydrates but relatively low in protein and fat. Mineral analysis revealed substantial variation among raw materials, with high concentrations of P, Mg, Ca, and K in several biomass sources. The formulated CFF contained 12.16 ± 0.21% crude protein, 3.29 ± 0.05% fat, and 47.56 ± 0.90% carbohydrates. The calculated EME content was 2685 kcal/kg of sample, which is highly comparable to the nutrient profiles of commercial feeds available in the market. Importantly, CFF production costs were markedly lower (4.50 THB/kg) than those of commercial products (10.67–15.33 THB/kg), demonstrating significant economic advantages. Overall, the results confirm that locally sourced biomass can be effectively used to formulate nutritionally adequate, mineral-rich, and low-cost fermented feed, thereby contributing to sustainable beef cattle production and reducing farmers' feed expenses.

Effects of Hydroponic Cultivation Systems on Growth Performance and Yield of Pakchoi (Brassica rapa var. chinensis)

Pakamas Thaenkaew — 2026-03-25

The purpose of this study was to investigate the effects of different hydroponic systems on the growth and yield of Pakchoi (Brassica rapa var. chinensis). The experiment was conducted using a Randomized Complete Block Design (RCBD) to compare three hydroponic cultivation systems: (1) Deep Flow Technique (DFT), (2) Nutrient Film Technique (NFT), and (3) Drip Hydroponics System (DHS). Each treatment consisted of 10 individual plants, yielding a total of 30 experimental units. Plants were cultivated for 35 days after transplanting under controlled laboratory conditions. The nutrient solution was maintained at a pH range of 5.5–6.5 and an electrical conductivity (EC) of 1.5–2.5 mS cm⁻¹. Environmental parameters were carefully regulated, including temperature (20–25 °C), relative humidity (60–70%), and light intensity (200–250 μmol m⁻² s⁻¹), under a 16 h light/8 h dark photoperiod. The results indicated that the cultivation system had a statistically significant effect (p < 0.05) on plant height, leaf width, and fresh weight yield. Pakchoi grown under the DHS treatment exhibited superior growth performance, with an average plant height of 41.71 cm, leaf width of 14.67 cm, and fresh weight yield of 64.8 g, all of which were higher than those observed in the NFT and DFT systems. In addition, qualitative observations revealed improved leaf expansion and greater canopy development under the DHS treatment. These findings demonstrate that the Drip Hydroponics System is an effective cultivation approach for enhancing Pakchoi growth and yield under controlled-environment conditions.

A Framework for a Geospatial Predictive Decision Support System for Hazard Risk Profiling

Anna Monica Paculaba — 2026-03-25

Natural hazards consistently pose significant challenges to urban communities, particularly in regions vulnerable to these threats, such as Catbalogan City, Samar, Philippines. Effective hazard risk profiling is essential for proactive disaster preparedness, resource allocation, and policy development. However, current approaches utilized by Catbalogan City and many Local Government Units (LGUs) in the Philippines continue to rely on static hazard maps and descriptive assessments, which limit predictive capability and decision-support effectiveness. This paper presents a conceptual, expert-validated framework for a Geospatial Predictive Decision Support System (GPDSS) to enhance hazard risk profiling and disaster preparedness. The study synthesizes findings from existing hazard-mapping, predictive modeling, and decision-support literature using a systematic review, a design science research approach, and a structured framework development process. The findings reveal persistent reliance on static data, regression-based analyses, and limited integration of decision-support mechanisms. These findings serve as the foundation for the proposed GPDSS framework, which integrates spatiotemporal geospatial datasets, GIS-based analysis, predictive modeling using machine learning techniques, and AI-assisted policy interpretation. Domain experts evaluated the framework, confirming its conceptual relevance, usability, scalability, and innovation for local disaster-risk governance. Rather than a deployed system, the proposed GPDSS is positioned as a governance-oriented framework that offers conceptual advantages over traditional hazard-mapping approaches by enabling predictive risk assessment, interactive visualization, and policy-oriented decision support. The study introduces a scalable, adaptive, and innovative framework intended to guide future system development and implementation for disaster-risk governance in hazard-prone localities in the Philippines.

Design and Development of a Smart Automated System for Poultry Farm Management Using Internet of Things (IOT)

Jaiinder Preet Singh — 2026-03-25

Worldwide, poultry farming is one of the fastest-growing sectors and is a key source of livelihood for thousands of small and medium-sized farmers in India, but maintaining optimal environmental conditions - temperature, humidity, and air quality - is critical and also difficult for the health and mortality of chicks. This research work leads to the design and development of a smart, automated poultry farm management system using the Internet of Things (IoT). The system consists of two major components: a hardware module built around a Raspberry Pi and a user-friendly mobile application. The hardware module includes environmental sensors: DHT22 for temperature and humidity, MQ137 for ammonia, and MQ135 for CO₂. The solution tracks environmental conditions and transmits all information in real time to the mobile app via Firebase Cloud. The mobile app allows farmers to remotely control all environmental parameters. It includes features that recommend rationing feed for future chicks based on age, record mortality rates, batch dates, breed, feed log, and sales records, and that provide access to all information via graphical apps for better visualization. It was observed to reduce chick mortality from 5% to 1.66%, which was a savings of approximately USD 34 (for every batch of 5000 chicks) and or a projected annual benefit of around USD 210. This solution has allowed farmers to be more productive and profitable while also reducing labour and stress during adverse weather conditions. It is worth noting that this success demonstrates the potential to transform traditional poultry farming into a smarter, more sustainable approach through the use of IoT.

Histology and Histochemistry of the Digestive Tract of Transverse-bar Barb, Hampala macrolepidota Kuhl & Van Hasselt, 1823

Akapon Vaniksampanna — 2026-03-25

Hampala macrolepidota is an important local fishery species; the information on its anatomy and microanatomy is scarce. Therefore, the histology and histochemistry of its digestive tract were systematically investigated. The digestive tracts of adult fish were removed and processed using the paraffin technique. Sections of 5 µm thickness were stained with hematoxylin and eosin (H&E), periodic acid-Schiff’s (PAS), and alcian blue (AB), pH 2.5 and pH 1.0. Its digestive tract consisted of the oropharynx, esophagus, and intestine, with no stomach observed. The stratified epithelium of the oropharynx was weakly positive-stained with PAS and AB at pH 2.5 and pH 1.0. In contrast, in the esophagus, mucous and goblet cells were strongly positively stained with all stains, indicating the secretion of neutral mucin and carboxylated and sulfated acid mucins. The intestine can be divided into three regions: anterior, middle, and posterior. Meanwhile, the enterocytes were positively stained with PAS; however, the goblet cells were positively stained with PAS at pH 2.5 and pH 1.0. The goblet cells secreted neutral mucins, as well as carboxylated and sulfated acidic mucins. The middle intestine showed crypt-like glands and the highest histometric values, indicating that it served as the main site of digestion and nutrient absorption, despite the absence of a stomach. The muscular layer varied regionally, with skeletal muscle in the oropharynx and esophagus and smooth muscle in the intestine. The intestinal coefficient was 0.69, indicating that it was a carnivorous fish. These findings provided fundamental data relevant to aquaculture, ecology, and comparative vertebrate anatomy.

Optimizing Brown Rice Drying With Modified LSU Systems For Energy Efficiency, Carbon Reduction, and 2AP Preservation

Sompot Khomkaew — 2026-03-25

This research examines how the quality of 2AP in Thai fragrant brown rice is affected by non-continuous LSU drying, emphasizing thermal energy management, moisture content, and aroma retention, which exhibit significant interrelationships and nonlinear responses across drying conditions and carbon emissions. The drying system employed an LSU-type dryer, a cyclone hot-air furnace, and a paddle mixer, using 4 tons of coal daily. This is equivalent to 26,680 kWh daily or an average power of 1,112 kW. The experiment took place in two stages: the first used 16,008 kWh/day to bring rice moisture down to 24.3-29.3%, averaging 26.5%, while the second used 10,672 kWh/day to reduce moisture to 10.9-12.4%, averaging 12.0%. During the 33.3% (stage-2 saving), energy use decreased the total system saving by 54.71%. In line with this, CO₂ emissions dropped from 13,337 kg.CO₂/day to 8,003 kg.CO₂/day a decrease of 5,334 kg.CO₂/day. Drying was meticulously controlled, with furnace temperatures maintained below 145°C and hot-air temperatures kept below 120°C. The research, conducted at a commercial rice mill operated by Maejaitana Chotiwat Co., Ltd., documented 2AP levels ranging from 1.83 to 2.67 µg/g, averaging 2.14 µg/g, with a mean drying temperature of 37.8°C. The results suggest that energy management, moisture control, and aroma retention are positively correlated. The study underscores the promise of low-emission, energy-efficient drying methods for improving product quality and promoting sustainability in the agro-industrial sector, with potential applications for crops such as cassava and corn.

A Hybrid Facial Expression Recognition System Based on Machine Learning and Deep Learning Models

Akkapop Prasompon — 2026-03-25

Pain assessment through facial expressions is an important area of research because many patients cannot clearly communicate their pain levels. This study presents the first systematic comparison of continuous time-series versus tokenized sequence representations for facial action unit (AU)-based pain classification, introducing a novel application of NLP models (BERT) to discretized AU sequences treated as symbolic text. Two datasets were used: the UNBC-McMaster Shoulder Pain Archive (UNBC-SP) with about 48,000 frames, and the Multimodal Intensity Pain (MIntPain) dataset with about 187,900 frames. Action unit intensities were extracted using the Py-Feat library and then normalized, oversampled, and augmented. A range of models was tested, including random forest, support vector machine, recurrent neural networks, and BERT. Key contributions include: (1) demonstrating that continuous time-series models significantly outperform tokenized approaches (91% vs. 82% accuracy); (2) revealing that classical ensemble methods surpass deep learning on tokenized sequences in data-limited scenarios; and (3) establishing that disruptive augmentations harm performance while conservative methods maintain accuracy. The continuous-time series models achieved the best performance, reaching 91% accuracy on MIntPain and 84% on UNBC-SP, while the tokenized models peaked at 82%. The results suggest that preserving temporal details of facial action units provides an advantage for pain detection, especially with larger datasets, though tokenization may retain value in resource-limited settings. The study highlights the need for larger, more diverse datasets and for validation in real clinical settings to improve the reliability of automatic pain recognition.