Pub Date : 2022-03-12DOI: 10.1109/ICIPRob54042.2022.9798717
YouweiI Wang, M. Mikawa, Makoto Fujisawa
Static environments are a prerequisite for most visual simultaneous localization and mapping (SLAM) systems because the dynamic matching points from moving objects in the camera’s field of view interrupt the localization process. The noise of the dynamic objects also contaminates the constructed maps. In this study, we propose a SLAM system designed to reduce the effects on the accuracy caused by dynamic objects to solve this issue. The noise points of dynamic objects are removed by combining depth information and semantic information. We evaluated the proposed method on the TUM RGB-D dataset, and the experimental results show that it performed well in dynamic environments, obtaining a high accuracy in most situations with a relatively high processing speed.
{"title":"FCH-SLAM: A SLAM Method for Dynamic Environments using Semantic Segmentation","authors":"YouweiI Wang, M. Mikawa, Makoto Fujisawa","doi":"10.1109/ICIPRob54042.2022.9798717","DOIUrl":"https://doi.org/10.1109/ICIPRob54042.2022.9798717","url":null,"abstract":"Static environments are a prerequisite for most visual simultaneous localization and mapping (SLAM) systems because the dynamic matching points from moving objects in the camera’s field of view interrupt the localization process. The noise of the dynamic objects also contaminates the constructed maps. In this study, we propose a SLAM system designed to reduce the effects on the accuracy caused by dynamic objects to solve this issue. The noise points of dynamic objects are removed by combining depth information and semantic information. We evaluated the proposed method on the TUM RGB-D dataset, and the experimental results show that it performed well in dynamic environments, obtaining a high accuracy in most situations with a relatively high processing speed.","PeriodicalId":435575,"journal":{"name":"2022 2nd International Conference on Image Processing and Robotics (ICIPRob)","volume":"43 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-03-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114873142","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 : 2022-03-12DOI: 10.1109/ICIPRob54042.2022.9798735
D. Mamoru, A. Panditha, W. Perera, G. U. Ganegoda
Playing video games has been popular across all the age limits of modern society. In the beginning, it was limited among the younger community and it was just a hobby limited to individuals. Even though the majority of society sees video gaming as having a negative impact on society, this modern industry acts a significant role in healing the present competitive, stressful society. Game commentary has played a major role in the domain of competitive ESports. A proper game commentary is beneficial to both players and the audience. The aim of this project is to analyze the gameplays to produce a commentary track while balancing the contributing factors, color commentary, and play-by-play commentary. The project consists of three modules that perform the study in three perspectives: 1. Word sets related to action, spatial, temporal, and statistical information, 2. Word sets related to color commentary, 3. Word sets related to play-by-play commentary. In each module, a game commentary is generated using only the word sets related to that module. For evaluation, the similarity between the human commentary and the generated commentaries individually will be calculated.
{"title":"Conceptual Representation and Evaluation of an FPS Game Commentary Generator","authors":"D. Mamoru, A. Panditha, W. Perera, G. U. Ganegoda","doi":"10.1109/ICIPRob54042.2022.9798735","DOIUrl":"https://doi.org/10.1109/ICIPRob54042.2022.9798735","url":null,"abstract":"Playing video games has been popular across all the age limits of modern society. In the beginning, it was limited among the younger community and it was just a hobby limited to individuals. Even though the majority of society sees video gaming as having a negative impact on society, this modern industry acts a significant role in healing the present competitive, stressful society. Game commentary has played a major role in the domain of competitive ESports. A proper game commentary is beneficial to both players and the audience. The aim of this project is to analyze the gameplays to produce a commentary track while balancing the contributing factors, color commentary, and play-by-play commentary. The project consists of three modules that perform the study in three perspectives: 1. Word sets related to action, spatial, temporal, and statistical information, 2. Word sets related to color commentary, 3. Word sets related to play-by-play commentary. In each module, a game commentary is generated using only the word sets related to that module. For evaluation, the similarity between the human commentary and the generated commentaries individually will be calculated.","PeriodicalId":435575,"journal":{"name":"2022 2nd International Conference on Image Processing and Robotics (ICIPRob)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-03-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130643741","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 : 2022-03-12DOI: 10.1109/ICIPRob54042.2022.9798728
H. Herath, K.G.C. Dulanga, N.V.D. Tharindu, G. U. Ganegoda
An authenticate users have increased due to failures in traditional authentication systems. Keystroke dynamics-based authentication is one of the most secure behavioral biometric authentication systems. This study aims to research and implement a non-fool proof, low-cost continuous authentication system for touch devices based on keystroke dynamics. A custom-developed mobile application was used to collect users’ keystroke dynamics. Bigrams were used as input parameters. 2 artificial neural networks were used in this study. The first network was used to identify users’ handedness, while the second one decided to use validity. Also, input was not limited, and users could type free text. As a result, overall accuracy was above 83.74%. Based on the results, we concluded that keystroke dynamics could be used for continuous user authentication purposes even with freely typed tests.
{"title":"Continuous User Authentication using Keystroke Dynamics for Touch Devices","authors":"H. Herath, K.G.C. Dulanga, N.V.D. Tharindu, G. U. Ganegoda","doi":"10.1109/ICIPRob54042.2022.9798728","DOIUrl":"https://doi.org/10.1109/ICIPRob54042.2022.9798728","url":null,"abstract":"An authenticate users have increased due to failures in traditional authentication systems. Keystroke dynamics-based authentication is one of the most secure behavioral biometric authentication systems. This study aims to research and implement a non-fool proof, low-cost continuous authentication system for touch devices based on keystroke dynamics. A custom-developed mobile application was used to collect users’ keystroke dynamics. Bigrams were used as input parameters. 2 artificial neural networks were used in this study. The first network was used to identify users’ handedness, while the second one decided to use validity. Also, input was not limited, and users could type free text. As a result, overall accuracy was above 83.74%. Based on the results, we concluded that keystroke dynamics could be used for continuous user authentication purposes even with freely typed tests.","PeriodicalId":435575,"journal":{"name":"2022 2nd International Conference on Image Processing and Robotics (ICIPRob)","volume":"104 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-03-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117199847","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 : 2022-03-12DOI: 10.1109/ICIPRob54042.2022.9798742
N. Premakumara, C. Shiranthika, Chathurangi Shyalika, Surangani Bandara
Text summarization is the task of condensing a text segment into a shorter version, reducing the size of the original text context while also preserving the informational elements and the meaning of the content. Manual text summarization will involve a significant amount of time and thus become a time expensive and generally laborious task. Aiming to reduce these pitfalls in manual text summarization, automatic text summarization has been evolving now bearing a strong motivation for academic research. Text Summarization is carried out by two main approaches, namely Extraction and Abstraction. This paper utilizes the extraction process for sentence selection. We also used some feature-based sentence scoring techniques, which play an important role in text summarization. Recently fuzzy logic-based research projects have been popularized among researchers and have been extensively applied in the domain of Natural Language Processing. Our main goal in this paper is to apply fuzzy logic in the task of text summarization. Finally, we analyzed the performance metrics resulting from the fuzzy logic-based text summarization with the benchmark methods; Rule Base and Neural Network techniques for computing the values for Precision, Recall, and F-Measure. In the process of applying the Fuzzy logic, rules were used to balance the weights between important and unimportant features based on the Feature Extraction. With the experimental results achieved, it was concluded that approaching Fuzzy Logic in the process of text summarization yields more successful results than the Rule Base and Neural Network methods.
{"title":"Optimized Text Summarization method based on Fuzzy Logic","authors":"N. Premakumara, C. Shiranthika, Chathurangi Shyalika, Surangani Bandara","doi":"10.1109/ICIPRob54042.2022.9798742","DOIUrl":"https://doi.org/10.1109/ICIPRob54042.2022.9798742","url":null,"abstract":"Text summarization is the task of condensing a text segment into a shorter version, reducing the size of the original text context while also preserving the informational elements and the meaning of the content. Manual text summarization will involve a significant amount of time and thus become a time expensive and generally laborious task. Aiming to reduce these pitfalls in manual text summarization, automatic text summarization has been evolving now bearing a strong motivation for academic research. Text Summarization is carried out by two main approaches, namely Extraction and Abstraction. This paper utilizes the extraction process for sentence selection. We also used some feature-based sentence scoring techniques, which play an important role in text summarization. Recently fuzzy logic-based research projects have been popularized among researchers and have been extensively applied in the domain of Natural Language Processing. Our main goal in this paper is to apply fuzzy logic in the task of text summarization. Finally, we analyzed the performance metrics resulting from the fuzzy logic-based text summarization with the benchmark methods; Rule Base and Neural Network techniques for computing the values for Precision, Recall, and F-Measure. In the process of applying the Fuzzy logic, rules were used to balance the weights between important and unimportant features based on the Feature Extraction. With the experimental results achieved, it was concluded that approaching Fuzzy Logic in the process of text summarization yields more successful results than the Rule Base and Neural Network methods.","PeriodicalId":435575,"journal":{"name":"2022 2nd International Conference on Image Processing and Robotics (ICIPRob)","volume":"191 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-03-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133276739","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 : 2022-03-12DOI: 10.1109/ICIPRob54042.2022.9798738
Yutaka Saragai, Takuya Sato, Haruki Kuroki, H. Ikeoka, Koichi Isawa
Sillago japonica is a popular fish used in Japanese cuisine, and a stable aquaculture method for growing fish of more than 25 cm in length, which are traded at high prices, may help to revitalize the aquaculture industry. However, it is difficult to raise Sillago japonica in aquaculture using conventional simple automatic feeding systems. Recently, AI and IoT have been used in aquaculture. Thus, we have been developing a fish distribution recognition system using image recognition AI: preprocessing of an automatic feeding control AI to optimize feeding timing. In this study, both fish positional recognition and fish directional recognition were achieved by recognizing the position of the head to fine-tune feeding using AI. Moreover, to adapt the system to the actual aquaculture environment, we implemented a mechanism to operate the actual feeder according to the instructions of the AI developed on the simulator. We mainly studied the following two items for practical use. First, we studied the mechanical operation of an actual feeder. Second, we built a wireless communication system between the feeding control and the feeder action processes.
{"title":"Study on Adapting the Auto Feeding System for Sillago Japonica to Actual Aquaculture Environment","authors":"Yutaka Saragai, Takuya Sato, Haruki Kuroki, H. Ikeoka, Koichi Isawa","doi":"10.1109/ICIPRob54042.2022.9798738","DOIUrl":"https://doi.org/10.1109/ICIPRob54042.2022.9798738","url":null,"abstract":"Sillago japonica is a popular fish used in Japanese cuisine, and a stable aquaculture method for growing fish of more than 25 cm in length, which are traded at high prices, may help to revitalize the aquaculture industry. However, it is difficult to raise Sillago japonica in aquaculture using conventional simple automatic feeding systems. Recently, AI and IoT have been used in aquaculture. Thus, we have been developing a fish distribution recognition system using image recognition AI: preprocessing of an automatic feeding control AI to optimize feeding timing. In this study, both fish positional recognition and fish directional recognition were achieved by recognizing the position of the head to fine-tune feeding using AI. Moreover, to adapt the system to the actual aquaculture environment, we implemented a mechanism to operate the actual feeder according to the instructions of the AI developed on the simulator. We mainly studied the following two items for practical use. First, we studied the mechanical operation of an actual feeder. Second, we built a wireless communication system between the feeding control and the feeder action processes.","PeriodicalId":435575,"journal":{"name":"2022 2nd International Conference on Image Processing and Robotics (ICIPRob)","volume":"30 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-03-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130200256","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 : 2022-03-12DOI: 10.1109/ICIPRob54042.2022.9798726
Toshinori Nagasawa, Kodai Oyake, M. Kawahara, S. Kaneko, G. Capi
In this paper, we present a new robotic system to assist elderly people during walking in outdoor environments. The walking assistive robot is equipped with depth camera, GPS, compass and ultrasonic sensors. The robot generates the route to the goal location using the google map. Based on the sensor information the robot follows the generated route while avoiding obstacles. In addition, the robot adjusts its speed based on the walking speed of the user and sends information to the family members of the user about the location in the google map. The robot checks the condition of the user and his/her distance to the robot and takes the appropriate decisions. The usefulness of the proposed system is examined experimentally.
{"title":"Development of a walking assistive robot for elderly people in outdoor environments","authors":"Toshinori Nagasawa, Kodai Oyake, M. Kawahara, S. Kaneko, G. Capi","doi":"10.1109/ICIPRob54042.2022.9798726","DOIUrl":"https://doi.org/10.1109/ICIPRob54042.2022.9798726","url":null,"abstract":"In this paper, we present a new robotic system to assist elderly people during walking in outdoor environments. The walking assistive robot is equipped with depth camera, GPS, compass and ultrasonic sensors. The robot generates the route to the goal location using the google map. Based on the sensor information the robot follows the generated route while avoiding obstacles. In addition, the robot adjusts its speed based on the walking speed of the user and sends information to the family members of the user about the location in the google map. The robot checks the condition of the user and his/her distance to the robot and takes the appropriate decisions. The usefulness of the proposed system is examined experimentally.","PeriodicalId":435575,"journal":{"name":"2022 2nd International Conference on Image Processing and Robotics (ICIPRob)","volume":"64 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-03-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122597295","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 : 2022-03-12DOI: 10.1109/ICIPRob54042.2022.9798719
P. M. Peiris, J. A. A. M. Jayaweera, G. U. Ganegoda
Melanoma is the deadliest form of skin cancer, whereas it has a metastases form when advanced into later stages. While skin cancers are most prominently seen in individuals with white skin, any individual can be diagnosed with skin cancers at any point in their life. Melanoma, mostly left untreated and undetected till its later stages make the patients’ lives be challenged, which has increased the importance of early-detecting. In this research, an effective approach is proposed for detecting Melanoma by analyzing gene DNA sequences of a subject, where the mutations are analyzed from nucleotide level up to the amino acid level. The research also consists of making sure the sequences are less fragmented when extracting, and also conducts a thorough analysis on the effect of various features such as gene, protein primary structure, age, tumor, tier, etc. to Melanoma with the help of machine learning algorithms. The obtained results are evaluated based on cross-validation and results from existing approaches.
{"title":"Melanoma Detection by Analysing Mutations in Gene DNA Sequences and Their Primary Protein Structures","authors":"P. M. Peiris, J. A. A. M. Jayaweera, G. U. Ganegoda","doi":"10.1109/ICIPRob54042.2022.9798719","DOIUrl":"https://doi.org/10.1109/ICIPRob54042.2022.9798719","url":null,"abstract":"Melanoma is the deadliest form of skin cancer, whereas it has a metastases form when advanced into later stages. While skin cancers are most prominently seen in individuals with white skin, any individual can be diagnosed with skin cancers at any point in their life. Melanoma, mostly left untreated and undetected till its later stages make the patients’ lives be challenged, which has increased the importance of early-detecting. In this research, an effective approach is proposed for detecting Melanoma by analyzing gene DNA sequences of a subject, where the mutations are analyzed from nucleotide level up to the amino acid level. The research also consists of making sure the sequences are less fragmented when extracting, and also conducts a thorough analysis on the effect of various features such as gene, protein primary structure, age, tumor, tier, etc. to Melanoma with the help of machine learning algorithms. The obtained results are evaluated based on cross-validation and results from existing approaches.","PeriodicalId":435575,"journal":{"name":"2022 2nd International Conference on Image Processing and Robotics (ICIPRob)","volume":"30 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-03-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130944788","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 : 2022-03-12DOI: 10.1109/ICIPRob54042.2022.9798734
Shun Oi, Kitahiro Kaneda, Keiichi Iwamura
Currently, supply chain management (SCM) manages products by externally attaching barcodes, RFID, etc. However, if RFID is physically peeled off from the product, it cannot be managed. As one of the solutions, SCM using physically unclonable function (PUF) is being considered, especially for devices such as semiconductor chips, genuine devices and counterfeit devices. An individual identification system has been devised that uses technology to identify the device. However, since PUF technology deals with minute physical features unique to the device, considerable difficulties remain in its practical application. Using a new authenticity judgment that overcomes the problems, we have realized and implemented SCM system using IoT device and blockchain that performs the same functions as PUF even for devices and substances for which PUF technology has not been established. We describe the evaluation and future of the implemented system.
{"title":"Implementation of Supply Chain Management System to Prevent Counterfeit Using IoT device and Blockchain","authors":"Shun Oi, Kitahiro Kaneda, Keiichi Iwamura","doi":"10.1109/ICIPRob54042.2022.9798734","DOIUrl":"https://doi.org/10.1109/ICIPRob54042.2022.9798734","url":null,"abstract":"Currently, supply chain management (SCM) manages products by externally attaching barcodes, RFID, etc. However, if RFID is physically peeled off from the product, it cannot be managed. As one of the solutions, SCM using physically unclonable function (PUF) is being considered, especially for devices such as semiconductor chips, genuine devices and counterfeit devices. An individual identification system has been devised that uses technology to identify the device. However, since PUF technology deals with minute physical features unique to the device, considerable difficulties remain in its practical application. Using a new authenticity judgment that overcomes the problems, we have realized and implemented SCM system using IoT device and blockchain that performs the same functions as PUF even for devices and substances for which PUF technology has not been established. We describe the evaluation and future of the implemented system.","PeriodicalId":435575,"journal":{"name":"2022 2nd International Conference on Image Processing and Robotics (ICIPRob)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-03-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131320010","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 : 2022-03-12DOI: 10.1109/ICIPRob54042.2022.9798733
Hayato Suzuki, Ryosuke Osawa, Hiroto Saegusa, S. Kaneko, G. Capi
With the growing world population, limited agriculture resources and reduced number of people working in agriculture sector, the need for intelligent autonomous robots is increasing. Agriculture robots must perform a wide range of operations like spraying pesticides, dispense fertilizers, remove weeds. In this paper, we explain in detail an agriculture robot developed in Assistive Robotics Laboratory, Hosei University. It consists of three subsystems: 1) The wheel-type actuated system; 2) A parallel link arm and 3) Fertilizer system. The robot utilizes the visual information and the Convolution Neural Networks to recognize the target plants. To evaluate the performance of developed robot, we performed experiments for spinach recognition, fertilizer dispenser and robot spraying.
{"title":"Development of Agriculture Robot for Plant Detection and Fertilizer Dispense","authors":"Hayato Suzuki, Ryosuke Osawa, Hiroto Saegusa, S. Kaneko, G. Capi","doi":"10.1109/ICIPRob54042.2022.9798733","DOIUrl":"https://doi.org/10.1109/ICIPRob54042.2022.9798733","url":null,"abstract":"With the growing world population, limited agriculture resources and reduced number of people working in agriculture sector, the need for intelligent autonomous robots is increasing. Agriculture robots must perform a wide range of operations like spraying pesticides, dispense fertilizers, remove weeds. In this paper, we explain in detail an agriculture robot developed in Assistive Robotics Laboratory, Hosei University. It consists of three subsystems: 1) The wheel-type actuated system; 2) A parallel link arm and 3) Fertilizer system. The robot utilizes the visual information and the Convolution Neural Networks to recognize the target plants. To evaluate the performance of developed robot, we performed experiments for spinach recognition, fertilizer dispenser and robot spraying.","PeriodicalId":435575,"journal":{"name":"2022 2nd International Conference on Image Processing and Robotics (ICIPRob)","volume":"44 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-03-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123895460","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 : 2022-03-12DOI: 10.1109/ICIPRob54042.2022.9798737
Yugo Kunisada, C. Premachandra
In this paper, we propose a generic learning method for training conditional generative adversarial networks on audio data. This makes it possible to apply the same generic approach as described in this study to problems that previously required completely different loss formulations when learning audio data. This method can be useful for labeling noises with a certain number of identical frequencies, generating speech labels corresponding to each frequency, and generating audio data for noise cancellation. To achieve this, we propose a sound restoration process based on U-Net, called Sound U-net. In this study, we realized a wide applicability of our system, owing to its ease of implementation without a parameter adjustment, as well as a reduction in the training time for audio data. During the experiment, reasonable results were obtained without manually adjusting the loss function.
{"title":"Sound-to-Sound Translation Using Generative Adversarial Network and Sound U-Net","authors":"Yugo Kunisada, C. Premachandra","doi":"10.1109/ICIPRob54042.2022.9798737","DOIUrl":"https://doi.org/10.1109/ICIPRob54042.2022.9798737","url":null,"abstract":"In this paper, we propose a generic learning method for training conditional generative adversarial networks on audio data. This makes it possible to apply the same generic approach as described in this study to problems that previously required completely different loss formulations when learning audio data. This method can be useful for labeling noises with a certain number of identical frequencies, generating speech labels corresponding to each frequency, and generating audio data for noise cancellation. To achieve this, we propose a sound restoration process based on U-Net, called Sound U-net. In this study, we realized a wide applicability of our system, owing to its ease of implementation without a parameter adjustment, as well as a reduction in the training time for audio data. During the experiment, reasonable results were obtained without manually adjusting the loss function.","PeriodicalId":435575,"journal":{"name":"2022 2nd International Conference on Image Processing and Robotics (ICIPRob)","volume":"25 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-03-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129358468","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：