Boonsang, Siridech

Preferred name

Boonsang, Siridech

Alternative Name

Boonsang, S.

Main Affiliation

School of Engineering

siridech.bo@kmitl.ac.th

Now showing 1 - 10 of 72

Some of the metrics are blocked by your
consent settings
Generative AI for Industrial Applications: Synthetic Dataset
(2023-01-01)
Sasiaowapak, Thanyathep
;
Boonsang, Siridech
;
Chuwongin, Santhad
;
Tongloy, Teerawat
;
Lalitrojwong, Pattarachai
The research and development of artificial intelligence (AI) techniques to enhance quality control in industrial equipment might face challenges due to the scarcity and limited privacy of actual industrial datasets. One approach to address this involves utilizing generative AI models that create synthetic data, simulating the characteristics and diversity found in crucial datasets. We present a methodology for generating synthetic datasets for industrial products such as bolts and screws by employing a segment-anything model and a stable diffusion technique for creating accurate representations. Furthermore, we propose the developed model by using a scaled-down version of DinoV2 algorithm's vision transformer (ViT-small). The self-supervised learning approach was studied to fine-tune the model to classify between normal- and defective industrial products, as well as those contaminated with dirt. By additional training dataset created through synthesis, we achieve an improvement in performance. The synthetic data leads to nearly perfect true positive results while completely eliminating false negatives. This indicates a significant advantage in terms of accuracy, recall, precision, specificity, and F1 score, all of which exceed 98%. Similarly, the model's predictions align perfectly with the area under the curve (AUC) metric. Although there is a slight performance reduction when dealing with up to six different class labels, the model retains strong capability in identifying normal products. Notably, the ViT-Small-based self-supervised learning model demonstrates superior accuracy compared to using ViT-Base, with considerations for dataset compatibility and model suitability. In conclusion, this study's contribution lies in enabling the deployment of the Dino V2 model for implementing quality control measures in industrial domains. It emphasizes the challenges that limited real-industrial data by leveraging synthetic data and innovative fine-tuning approaches, ultimately enhancing AI-powered for quality control processes.
Some of the metrics are blocked by your
consent settings
A highly sensitive laser-EMAT imaging system for biomedical applications
(2014-10-15)
Boonsang, S.
;
Dewhurst, R. J.
This paper describes a novel technique using an elctromagnetic acoustic transducer (EMAT) for biomedical photoacoustic measurements. The physical principle underlying and EMAT sensor is based on the detection of electromagnetic signals, which are responses from the interaction of ultrasonic waves and magnetic fields within the electrically conductive sample surface. A recent prototype EMAT sensor has miniaturized the receiver size (1.0 cm diameter). It is incorporated with a specially designed low noise preamplifier. Calibration procedures with a Michelson interferometer revealed that the EMAT sensor detected small displacement amplitudes as low as 1.0±0.2 pm in ultrasonic frequency range using an aluminium sample. For the first time, we present the use of an EMAT sensor to detect photoacoustic pressure signals in tissue phantoms and real tissues (chicken breast) in vitro. Corresponding B-scan images were constructed by the compilation of a series of received photoacoustic signals. A B-scan image of human hairs in a chicken breast sample revealed that the lateral resolution about 1.57 mm (FWHM) was achieved.
Some of the metrics are blocked by your
consent settings
Cloud-based optical characterization of hard disk drive recording heads
(2018-01-01)
Khunrattanasiri, Weerayuth
;
Yokpradit, Anchyza
;
Tongloy, Teerawat
;
Boonsang, Siridech
In the hard disk drive (HDD) industry, the measurement of flying height is crucial for the verification of the recording head air-bearing design. The flying performance of recording heads has to be measured and compared with the original design to assure the reliability of HDDs. Optical interferometry is normally used to measure the flying height of recording heads. The key parameter for setting up the measurement is the refractive index, whose values normally deviate depending on machine conditions during the manufacturing processes. The variation in refractive index cannot be identified during flying height measurement because there is no on-line technique for doing so. In this paper, we present a novel technique using inhouse designed micro-ellipsometry to measure the complex refractive index combined with data transfer to a private cloud network via the Internet of Things message queuing telemetry transport (IoT MQTT) protocol. Therefore, the measurement of the complex refractive index can be carried out for each recording head and we can improve the precision of the flying height measurement. In addition, the information about refractive index not only can be shared with a flying height tester to correct the variation in each recording head but also can be used to assist air-bearing designers in the instantaneous validation of their design.
Some of the metrics are blocked by your
consent settings
Automatic identification of medically important mosquitoes using embedded learning approach-based image-retrieval system
(2023-12-01)
Kittichai, Veerayuth
;
Kaewthamasorn, Morakot
;
Samung, Yudthana
;
Jomtarak, Rangsan
;
Naing, Kaung Myat
Mosquito-borne diseases such as dengue fever and malaria are the top 10 leading causes of death in low-income countries. Control measure for the mosquito population plays an essential role in the fight against the disease. Currently, several intervention strategies; chemical-, biological-, mechanical- and environmental methods remain under development and need further improvement in their effectiveness. Although, a conventional entomological surveillance, required a microscope and taxonomic key for identification by professionals, is a key strategy to evaluate the population growth of these mosquitoes, these techniques are tedious, time-consuming, labor-intensive, and reliant on skillful and well-trained personnel. Here, we proposed an automatic screening, namely the deep metric learning approach and its inference under the image-retrieval process with Euclidean distance-based similarity. We aimed to develop the optimized model to find suitable miners and suggested the robustness of the proposed model by evaluating it with unseen data under a 20-returned image system. During the model development, well-trained ResNet34 are outstanding and no performance difference when comparing five data miners that showed up to 98% in its precision even after testing the model with both image sources: stereomicroscope and mobile phone cameras. The robustness of the proposed—trained model was tested with secondary unseen data which showed different environmental factors such as lighting, image scales, background colors and zoom levels. Nevertheless, our proposed neural network still has great performance with greater than 95% for sensitivity and precision, respectively. Also, the area under the ROC curve given the learning system seems to be practical and empirical with its value greater than 0.960. The results of the study may be used by public health authorities to locate mosquito vectors nearby. If used in the field, our research tool in particular is believed to accurately represent a real-world scenario.
Some of the metrics are blocked by your
consent settings
Highly-transparent multi-layered spin-coated silk fibroin film
(2017-01-01)
Wasapinyokul, Kamol
;
Kaewpirom, Supranee
;
Chuwongin, Santhad
;
Boonsang, Siridech
In this study, the silk fibroin films with different numbers of layers were fabricated by the spin-coating method and their optical transmittances were observed. The process to synthesise the silk fibroin solution was explained - starting from the silk cocoon until the silk-fibroin solution, approximately 7.5% concentration wt/vol, was obtained. The solution was spin-coated onto clean glass substrates to fabricate samples. Totally 10 samples with different numbers of layers, from 1 to 5 layers, were obtained. All samples can be separated into two groups: those left dried at room temperature after spin-coating and those heated at 60°C. They were then measured for their transmittance over the visible-to-near-infrared region. All samples exhibited the high transmittance where the values were at 95% and 98%, for the samples at room temperature and those at 60°C, respectively. This was believed to be due to the heating effect that caused the silk fibroin to arrange itself after being heated, hence the higher transmittance. These high transmittances were maintained regardless of the number of layers and length of heating time. Results from this study could be used to fabricate a silk fibroin film with high optical transmittance and adjustable other properties.
Some of the metrics are blocked by your
consent settings
Artificial intelligence assistance in radiographic detection and classification of knee osteoarthritis and its severity: A cross-sectional diagnostic study
(2022-01-01)
Pongsakonpruttikul, N.
;
Angthong, Chayanin
;
Kittichai, V.
;
Chuwongin, S.
;
Puengpipattrakul, P.
OBJECTIVE: Radiographic interpretation suffers from an ever-increasing workload in orthopedic and radiology departments. The present study applied and assessed the performance of a convolutional neural network designed to assist orthopedists and radiologists in the detection and classification of knee osteoarthritis from early to severe degrees in accordance with the Kellgren-Lawrence (KL) classification system. MATERIALS AND METHODS: In total, 1650 knee joint radiographs (anteroposterior view) were collected from the Osteoarthritis Initiative public resource. Two models were developed: one distinguished normal (KL 0-I) from osteoarthritic knees (KL II-IV), and the other classified the severity as normal (KL 0-I), non-severe (KL II), or severe (KL III-IV). The regions of interest were labeled under the supervision of experts. Our artificial intelligence (AI) models were trained using the You Only Look Once version 3 (YOLOv3) detection algorithm. RESULTS: Our first AI model using YOLOv3 tiny could detect and classify normal and osteoarthritic knees on plain knee joint radiographs with 85% accuracy and 81% mean average precision. The second AI model for classifying severity achieved a total accuracy of 86.7% and mean average precision of 70.6%. CONCLUSIONS: Our proposed deep learning models provided high accuracy and satisfactory precision for the detection and classification of early to severe knee osteoarthritis on anteroposterior radiographs. These models may be used as diagnostic aids by interpreting knee radiographs and guiding the treatment options via each osteoarthritic stage for related physicians and specialists.
Some of the metrics are blocked by your
consent settings
Uncertainty analysis of polarization interferometric measurements of hard disk drive recording head-media dynamic spacing
(2012-06-12)
Boonsang, Siridech
;
Aroonjarernchay, Wanchai
In this paper the numerical method based on Monte-Carlo techniques is utilized for the analysis of the flying height measurement uncertainty. In the current flying height tester, the prediction of error is the main concern in order to assure the attainable precision. In previous study, a simplified analytical method is normally used and it may not provide the accurate results of the error prediction especially when the amount of the flying height values becoming as low as sub-5 nm. In this paper, instead of using the analytical method for error analysis, the numerical method based on Monte-Carlo techniques is utilized. Our simulation results show that the measurement uncertainty is mainly from the variation in the optical constants of the DLC layer. They also show that the increasing of standard deviation of the optical constants of DLC layer causes the higher measurement uncertainty.
Some of the metrics are blocked by your
consent settings
Intelligent identification of medical and veterinary intracellular protozoa by using self-supervised learning
(2026-12-01)
Kittichai, Veerayuth
;
Kaewthamasorn, Morakot
;
Soiphet, Naruchit
;
Tongloy, Teerawat
;
Chuwongin, Santhad
Background: Zoonotic diseases pose a major threat to both human and animal health, contributing significantly to global morbidity and mortality. Accurate diagnosis is crucial for effective control and treatment, with microscopic examination serving as the gold standard, supplemented by highly sensitive molecular biology techniques. However, these confirmatory methods require skilled personnel and are subject to inter- and intra-rater variability. An innovative solution lies in artificial intelligence (AI)-powered automated tools, which offer a promising alternative. This study aimed to develop a self-supervised learning (SSL) approach using the Distillation with No Labels (DiNOv2) algorithm to extract features of protozoa from Giemsa-stained blood samples. Methods: The development of self-supervised learning algorithms, including DiNOv2, was based on archived samples of clinically significant and significant veterinary microorganisms. These models were evaluated in comparison to a baseline Vision Transformer (ViT). Results: Among the tested SSL models, the DiNOv2-Small version achieved exceptional performance, surpassing 99% accuracy and specificity while maintaining the lowest misclassification rate (0.263). It also demonstrated a high area under the curve (AUC) value of 0.990, underscoring its robust classification capability. Remarkably, even when trained on only 20% of the dataset, the SSL models retained performance levels comparable to baseline models trained on the full dataset. However, further reducing the sample size below 20% led to notable declines in evaluation metrics: accuracy dropped by 5–7%, recall decreased from 37.9% to 35.1%, precision fell from 35% to 25.2%, and the F1 score declined from 31.1% to 21.5%. Additionally, the AUC decreased by 4–11%, while the misclassification rate increased, indicating reduced robustness. A key limitation of this study was the highly imbalanced fine-tuned dataset between classes. Nevertheless, the inconsistent performance observed may be mitigated by employing the larger DiNOv2 model, which improves the F1 score and enhances the model’s ability to handle imbalanced data while reducing reliance on labeled data. Conclusions: The proposed method can assist laboratory technicians, particularly in resource-limited healthcare settings. Furthermore, the findings support the potential deployment of this AI-based tool for automated screening in both medical and veterinary applications.
Some of the metrics are blocked by your
consent settings
Classification for avian malaria parasite Plasmodium gallinaceum blood stages by using deep convolutional neural networks
(2021-12-01)
Kittichai, Veerayuth
;
Kaewthamasorn, Morakot
;
Thanee, Suchansa
;
Jomtarak, Rangsan
;
Klanboot, Kamonpob
The infection of an avian malaria parasite (Plasmodium gallinaceum) in domestic chickens presents a major threat to the poultry industry because it causes economic loss in both the quality and quantity of meat and egg production. Computer-aided diagnosis has been developed to automatically identify avian malaria infections and classify the blood infection stage development. In this study, four types of deep convolutional neural networks, namely Darknet, Darknet19, Darknet19-448 and Densenet201 are used to classify P. gallinaceum blood stages. We randomly collected a dataset of 12,761 single-cell images consisting of three parasite stages from ten-infected blood films stained by Giemsa. All images were confirmed by three well-trained examiners. The study mainly compared several image classification models and used both qualitative and quantitative data for the evaluation of the proposed models. In the model-wise comparison, the four neural network models gave us high values with a mean average accuracy of at least 97%. The Darknet can reproduce a superior performance in the classification of the P. gallinaceum development stages across any other model architectures. Furthermore, the Darknet has the best performance in multiple class-wise classification, with average values of greater than 99% in accuracy, specificity, and sensitivity. It also has a low misclassification rate (< 1%) than the other three models. Therefore, the model is more suitable in the classification of P. gallinaceum blood stages. The findings could help us create a fast-screening method to help non-experts in field studies where there is a lack of specialized instruments for avian malaria diagnostics.
Some of the metrics are blocked by your
consent settings
Development of Self-Supervised Learning with Dinov2-Distilled Models for Parasite Classification in Screening
(2023-01-01)
Pinetsuksai, Natchapon
;
Kittichai, Veerayuth
;
Jomtarak, Rangsan
;
Jaksukam, Komgrit
;
Tongloy, Teerawat
At present, parasitic infections in humans, such as intestinal parasitic infections and soil-transmitted helminth (STH) infection, remain a public health concern, with screening methods that are simple but time-consuming and require parasitology experts. Microscopy images are increasingly being used to aid diagnosis but creating labels for supervised learning (SL) is a time-consuming, labor-intensive, and costly process. Self-supervised learning (SSL) is a deep learning approach that aims to train models to represent features in unlabeled datasets using automatically generated labels or annotations from the data itself, rather than explicitly labeled human-labeled labels. It is an appropriate method to address the challenges associated with the difficulty of labeling large datasets. A pretrained model that has learned useful data representations from an SSL task is fine-tuned using labeled data to perform well on a specific downstream task. DINOv2 is an SSL model based on the Vision Transformer (ViT) architecture. In this study, we aim to create a model for screening for helminth egg infection using a fine-tuned Dinov2 with a classification layer head to demonstrate that dataset sizes of 1% and 10% are sufficient when compared to SL model. Rather than SL, which requires a significant amount of human data labeling and is generally impractical, the model developed in this study is expected to be used in active surveillance in the future.

Boonsang, Siridech

Filters

Settings

Sort By

Results per page