Pub Date : 2024-02-16DOI: 10.1109/RESTCON60981.2024.10463558
Rolando Pula, Lorena Ilagan, Marcelo Santos
This study explores various techniques for transforming 1-dimensional time-series data into 2-dimensional images, preparing for the application of machine learning models designed for 2D data. Eight distinct methods are introduced, including recurrence plots, Markov transition, Gramian angular field, spectrogram, heatmap, direct plot, phase space transformation, and Poincaré plots. These methods are tested using data from a modeled photovoltaic (PV) grid-connected system, specifically simulating a shorted string fault and a no-fault condition. The fault and no-fault responses are captured with a fixed window size of 256 sample points, consistently applied across all methods. All transformation method is tested through python 3 programming using a laptop with minimal computing capability. The generated image of each transformation may contain 1-channel image in grayscale or 3-channel RGB image. Dimension of the generated image can be increase or decrease during saving process. Each method produces a unique visual representation of the shorted string fault and a no-fault, demonstrating diverse perspectives in transforming 1D time-series data into 2D images for subsequent machine learning applications.
{"title":"Methods of Signal to Image Transformation in Photovoltaic Fault Diagnosis in Preparation for Machine Learning Applications","authors":"Rolando Pula, Lorena Ilagan, Marcelo Santos","doi":"10.1109/RESTCON60981.2024.10463558","DOIUrl":"https://doi.org/10.1109/RESTCON60981.2024.10463558","url":null,"abstract":"This study explores various techniques for transforming 1-dimensional time-series data into 2-dimensional images, preparing for the application of machine learning models designed for 2D data. Eight distinct methods are introduced, including recurrence plots, Markov transition, Gramian angular field, spectrogram, heatmap, direct plot, phase space transformation, and Poincaré plots. These methods are tested using data from a modeled photovoltaic (PV) grid-connected system, specifically simulating a shorted string fault and a no-fault condition. The fault and no-fault responses are captured with a fixed window size of 256 sample points, consistently applied across all methods. All transformation method is tested through python 3 programming using a laptop with minimal computing capability. The generated image of each transformation may contain 1-channel image in grayscale or 3-channel RGB image. Dimension of the generated image can be increase or decrease during saving process. Each method produces a unique visual representation of the shorted string fault and a no-fault, demonstrating diverse perspectives in transforming 1D time-series data into 2D images for subsequent machine learning applications.","PeriodicalId":518254,"journal":{"name":"2024 1st International Conference on Robotics, Engineering, Science, and Technology (RESTCON)","volume":"81 11","pages":"195-200"},"PeriodicalIF":0.0,"publicationDate":"2024-02-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140527540","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-02-16DOI: 10.1109/RESTCON60981.2024.10463586
Keita Nakano, Kousuke Matsushima
In Advanced Driver-Assistance Systems (ADAS) and automatic driving, it is important to accurately recognize objects around the vehicle. DETReg is one of the unsupervised pre-training methods using Transformer, which is self-supervised by combining localization and categorization. DETReg performs self-supervised learning on unlabeled images. Then, it extracted a wide range of features from rich aspects of the data and gained the flexibility to adapt to many variations. Fine tuning then used the labeled dataset of the target task to fine tune the model to fit the specific dataset. This allowed DETReg to achieve higher accuracy in the object detection task. However, it is difficult to learn DETReg efficiently because of its slow learning time. In this paper, we propose a new pre-training method for object detection, called Semi-DETReg, that utilizes a few supervised labels during self-supervised learning. We incorporate semi-supervised learning into DETReg by using a portion of the supervised training data in the pre-training to improve efficiency. We demonstrate the effectiveness of our method by conducting experiments and comparing our method to a similarly trained DETReg.
{"title":"DETReg Incorporating Semi-Supervised Learning for Object Detection in the Advanced Driver-Assistance Systems","authors":"Keita Nakano, Kousuke Matsushima","doi":"10.1109/RESTCON60981.2024.10463586","DOIUrl":"https://doi.org/10.1109/RESTCON60981.2024.10463586","url":null,"abstract":"In Advanced Driver-Assistance Systems (ADAS) and automatic driving, it is important to accurately recognize objects around the vehicle. DETReg is one of the unsupervised pre-training methods using Transformer, which is self-supervised by combining localization and categorization. DETReg performs self-supervised learning on unlabeled images. Then, it extracted a wide range of features from rich aspects of the data and gained the flexibility to adapt to many variations. Fine tuning then used the labeled dataset of the target task to fine tune the model to fit the specific dataset. This allowed DETReg to achieve higher accuracy in the object detection task. However, it is difficult to learn DETReg efficiently because of its slow learning time. In this paper, we propose a new pre-training method for object detection, called Semi-DETReg, that utilizes a few supervised labels during self-supervised learning. We incorporate semi-supervised learning into DETReg by using a portion of the supervised training data in the pre-training to improve efficiency. We demonstrate the effectiveness of our method by conducting experiments and comparing our method to a similarly trained DETReg.","PeriodicalId":518254,"journal":{"name":"2024 1st International Conference on Robotics, Engineering, Science, and Technology (RESTCON)","volume":"80 8","pages":"123-128"},"PeriodicalIF":0.0,"publicationDate":"2024-02-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140527541","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-02-16DOI: 10.1109/restcon60981.2024.10463575
{"title":"RESTCON 2024 Messages and Keynote Speakers","authors":"","doi":"10.1109/restcon60981.2024.10463575","DOIUrl":"https://doi.org/10.1109/restcon60981.2024.10463575","url":null,"abstract":"","PeriodicalId":518254,"journal":{"name":"2024 1st International Conference on Robotics, Engineering, Science, and Technology (RESTCON)","volume":"537 ","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-02-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140527932","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-02-16DOI: 10.1109/RESTCON60981.2024.10463592
Torsak Boonthai, P. Nunthavarawong, P. Kowitwarangkul, Masaki Fuchiwaki
Mechanical alloying plays a crucial role in controlling and enhancing the characteristics of powder materials which in turn influence the quality and performance of thermal spray coatings. In this study, the optimal milling parameters for mechanical alloying were determined, specifically a rotational speed of 60 rpm, a milling period of 6 hours, and wet milling conditions. These settings led to the best preparation of feedstock powder, resulting in a minimal particle size of 17.5 µm and a narrow particle size dispersion. The observed extensive cataracting and impact zones at this rotational speed of 60 rpm corresponded to 65 of the % critical speed, indicating enhanced milling efficiency. Additionally, DEM modeling demonstrated good agreement with experimental findings, indicating enhanced milling efficiency. Additionally, DEM modeling demonstrated good agreement with experimental findings, highlighting that this rotational speed induced a cataracting regime characterized by a broad zone of impacted particles, yielding the highest impact velocity of 1.79 m/s and a ball indenter force interaction of 1.19 N.
{"title":"Mechanical Alloying Process Design by Using DEM Simulation and Experimental Validation","authors":"Torsak Boonthai, P. Nunthavarawong, P. Kowitwarangkul, Masaki Fuchiwaki","doi":"10.1109/RESTCON60981.2024.10463592","DOIUrl":"https://doi.org/10.1109/RESTCON60981.2024.10463592","url":null,"abstract":"Mechanical alloying plays a crucial role in controlling and enhancing the characteristics of powder materials which in turn influence the quality and performance of thermal spray coatings. In this study, the optimal milling parameters for mechanical alloying were determined, specifically a rotational speed of 60 rpm, a milling period of 6 hours, and wet milling conditions. These settings led to the best preparation of feedstock powder, resulting in a minimal particle size of 17.5 µm and a narrow particle size dispersion. The observed extensive cataracting and impact zones at this rotational speed of 60 rpm corresponded to 65 of the % critical speed, indicating enhanced milling efficiency. Additionally, DEM modeling demonstrated good agreement with experimental findings, indicating enhanced milling efficiency. Additionally, DEM modeling demonstrated good agreement with experimental findings, highlighting that this rotational speed induced a cataracting regime characterized by a broad zone of impacted particles, yielding the highest impact velocity of 1.79 m/s and a ball indenter force interaction of 1.19 N.","PeriodicalId":518254,"journal":{"name":"2024 1st International Conference on Robotics, Engineering, Science, and Technology (RESTCON)","volume":"424 1","pages":"133-138"},"PeriodicalIF":0.0,"publicationDate":"2024-02-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140527938","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-02-16DOI: 10.1109/RESTCON60981.2024.10463579
Korawee Hirunthakingpunt, Don Dawan, Chikamune Wada, Natinun Maneerung
Mobile robots are widely used in many departments such as industry, hospitals, restaurants, etc. The human detection and the human pose classification are usually used for human-robot interaction. The current study proposes human pose classification for human-robot interaction to avoid the collision. There are three main steps of the presented method. First, the algorithm detects the entire human within the determined range of 3D camera. Second, the K-Nearest Neighbor (KNN) model with skeleton points features is used for classifying the six postures of detected human such as neutral, left and right raise, both hand raise, cross hand posture and one opening hand forward. According to the posture classification, these can command the robot to move forward, stop, stop for a few seconds, and cancel the command. Finally, the command is used to interacting with the mobile robot to control the robot movement and to avoid the collision. The experiment results show that the designed algorithm can effectively detect and classify human posture with 86.14% for the accuracy of algorithms and interact effectively to avoid the collisions stop automatically within the 1.8 meters between human and robot.
{"title":"Human Detection and Human Pose Classification for Mobile Robots Interaction","authors":"Korawee Hirunthakingpunt, Don Dawan, Chikamune Wada, Natinun Maneerung","doi":"10.1109/RESTCON60981.2024.10463579","DOIUrl":"https://doi.org/10.1109/RESTCON60981.2024.10463579","url":null,"abstract":"Mobile robots are widely used in many departments such as industry, hospitals, restaurants, etc. The human detection and the human pose classification are usually used for human-robot interaction. The current study proposes human pose classification for human-robot interaction to avoid the collision. There are three main steps of the presented method. First, the algorithm detects the entire human within the determined range of 3D camera. Second, the K-Nearest Neighbor (KNN) model with skeleton points features is used for classifying the six postures of detected human such as neutral, left and right raise, both hand raise, cross hand posture and one opening hand forward. According to the posture classification, these can command the robot to move forward, stop, stop for a few seconds, and cancel the command. Finally, the command is used to interacting with the mobile robot to control the robot movement and to avoid the collision. The experiment results show that the designed algorithm can effectively detect and classify human posture with 86.14% for the accuracy of algorithms and interact effectively to avoid the collisions stop automatically within the 1.8 meters between human and robot.","PeriodicalId":518254,"journal":{"name":"2024 1st International Conference on Robotics, Engineering, Science, and Technology (RESTCON)","volume":"455 4","pages":"1-6"},"PeriodicalIF":0.0,"publicationDate":"2024-02-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140527936","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-02-16DOI: 10.1109/RESTCON60981.2024.10463583
Natthapol Saovana, Chavanont Khosakitchalert
Construction scaffolding serves as a pivotal temporary structure essential for construction activities, exerting a direct influence on site safety conditions. Unfortunately, the lack of documentation often leads to a shortage of training data necessary for employing image segmentation through deep learning for inspection purposes. In an effort to overcome this bottleneck, Generative AI, adept at creating images from pretrained data, emerges as a potential solution. However, the inherent black box nature of deep learning introduces the possibility of generating unrealistic images, thereb necessitating a rigorous evaluation, which constitutes the primary focus of our research. Our findings reveal that scaffolding images generated by Generative AI exhibit distinct features that our deep learning model successfully learned, resulting in an impressive mean average precision (mAP) of 82. Nonetheless, discernible patterns in image generation may be lacking, as evidenced by our deep learning system's ability to grasp scaffolding features proficiently, achieving a mAP of 69 even from the initial epoch. This observation suggests potential challenges in generating diverse scaffolding images through the Generative AI approach, emphasizing the need for further investigation before implementing it with real scenario images
{"title":"Assessing the Viability of Generative AI-Created Construction Scaffolding for Deep Learning-Based Image Segmentation","authors":"Natthapol Saovana, Chavanont Khosakitchalert","doi":"10.1109/RESTCON60981.2024.10463583","DOIUrl":"https://doi.org/10.1109/RESTCON60981.2024.10463583","url":null,"abstract":"Construction scaffolding serves as a pivotal temporary structure essential for construction activities, exerting a direct influence on site safety conditions. Unfortunately, the lack of documentation often leads to a shortage of training data necessary for employing image segmentation through deep learning for inspection purposes. In an effort to overcome this bottleneck, Generative AI, adept at creating images from pretrained data, emerges as a potential solution. However, the inherent black box nature of deep learning introduces the possibility of generating unrealistic images, thereb necessitating a rigorous evaluation, which constitutes the primary focus of our research. Our findings reveal that scaffolding images generated by Generative AI exhibit distinct features that our deep learning model successfully learned, resulting in an impressive mean average precision (mAP) of 82. Nonetheless, discernible patterns in image generation may be lacking, as evidenced by our deep learning system's ability to grasp scaffolding features proficiently, achieving a mAP of 69 even from the initial epoch. This observation suggests potential challenges in generating diverse scaffolding images through the Generative AI approach, emphasizing the need for further investigation before implementing it with real scenario images","PeriodicalId":518254,"journal":{"name":"2024 1st International Conference on Robotics, Engineering, Science, and Technology (RESTCON)","volume":"56 8","pages":"38-43"},"PeriodicalIF":0.0,"publicationDate":"2024-02-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140527546","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-02-16DOI: 10.1109/RESTCON60981.2024.10463567
Panas Thongtaweechaikij, Piyawat Tangpong, J. Inthiam, W. Tangsuksant
This study evaluates Optical Character Recognition's (OCR) effectiveness in extracting and organizing data from student cards. Assessing diverse OCR techniques, it aims to identify optimal methods for accurate text extraction, considering different formats and languages. The research investigates OCR's impact on information retrieval, analyzing its integration into databases for improved searchability and usability. Our proposed method presents the pre-processing with OCR process including the SIFT, KNN feature matching, MSER technique for noise detection and image transformation. For the experiment, all student cards in King Mongkut’s University of Technology North Bangkok capturing by smartphone, which the resolution of camera is greater than 2 megapixel. This research compares the different technique between traditional tesseract OCR and our proposed method by setting 50% and 70% of Intersection over Union (IoU), The experiment result shows that our proposed method with 70% of IoU has the highest accuracy as 97.36%. According to the result, the proposed illustrate the feasible method for our system.
{"title":"Text Extraction by Optical Character Recognition-Based on the Template Card","authors":"Panas Thongtaweechaikij, Piyawat Tangpong, J. Inthiam, W. Tangsuksant","doi":"10.1109/RESTCON60981.2024.10463567","DOIUrl":"https://doi.org/10.1109/RESTCON60981.2024.10463567","url":null,"abstract":"This study evaluates Optical Character Recognition's (OCR) effectiveness in extracting and organizing data from student cards. Assessing diverse OCR techniques, it aims to identify optimal methods for accurate text extraction, considering different formats and languages. The research investigates OCR's impact on information retrieval, analyzing its integration into databases for improved searchability and usability. Our proposed method presents the pre-processing with OCR process including the SIFT, KNN feature matching, MSER technique for noise detection and image transformation. For the experiment, all student cards in King Mongkut’s University of Technology North Bangkok capturing by smartphone, which the resolution of camera is greater than 2 megapixel. This research compares the different technique between traditional tesseract OCR and our proposed method by setting 50% and 70% of Intersection over Union (IoU), The experiment result shows that our proposed method with 70% of IoU has the highest accuracy as 97.36%. According to the result, the proposed illustrate the feasible method for our system.","PeriodicalId":518254,"journal":{"name":"2024 1st International Conference on Robotics, Engineering, Science, and Technology (RESTCON)","volume":"169 3","pages":"188-192"},"PeriodicalIF":0.0,"publicationDate":"2024-02-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140527944","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：