A. A. Puzi, S. N. Sidek, I. M. Khairuddin, H. Yusof
An important component in rehabilitation process is the ability to assess the level of muscle spasticity in objective manner. Despite of many proven evidences, current method of assessments is still based on subjective evaluation which relies heavily on the skill, experience and intuition of the therapists. Thus, this paper aims to develop a classifier of muscle spasticity level based on the clinical data collected from the affected upper limb. In order to quantify the assessment systematically, a standard Modified Ashworth Scale (MAS) tool was used to help develop ANFIS and SVM models. Data were collected from twenty-five subjects that met the requirements with prior consent. The data went through preprocessing stage and analyzed before seven features from the data were selected to form the dataset. The features were then went through one way ANOVA test to evaluate their statistically significant differences so as to best predict the assessment levels. Based on the results from the analysis, four features were finally selected and were then used to train the classifiers. The overall results from the ANFIS and SVM classifiers accorded the performance of 53.3% and 88.0% accuracy respectively in classifying the level of muscle spasticity.
{"title":"Objective Assessment for Classification of Muscle Spasticity Level","authors":"A. A. Puzi, S. N. Sidek, I. M. Khairuddin, H. Yusof","doi":"10.1145/3440084.3441181","DOIUrl":"https://doi.org/10.1145/3440084.3441181","url":null,"abstract":"An important component in rehabilitation process is the ability to assess the level of muscle spasticity in objective manner. Despite of many proven evidences, current method of assessments is still based on subjective evaluation which relies heavily on the skill, experience and intuition of the therapists. Thus, this paper aims to develop a classifier of muscle spasticity level based on the clinical data collected from the affected upper limb. In order to quantify the assessment systematically, a standard Modified Ashworth Scale (MAS) tool was used to help develop ANFIS and SVM models. Data were collected from twenty-five subjects that met the requirements with prior consent. The data went through preprocessing stage and analyzed before seven features from the data were selected to form the dataset. The features were then went through one way ANOVA test to evaluate their statistically significant differences so as to best predict the assessment levels. Based on the results from the analysis, four features were finally selected and were then used to train the classifiers. The overall results from the ANFIS and SVM classifiers accorded the performance of 53.3% and 88.0% accuracy respectively in classifying the level of muscle spasticity.","PeriodicalId":250100,"journal":{"name":"Proceedings of the 2020 4th International Symposium on Computer Science and Intelligent Control","volume":"32 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-11-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125157168","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}
Behavior tree (BT) is a novel control architecture in the robotic field. Being modular and reactive, BTs show great advantages in autonomous systems when applied to robot control. However, there are unsolved problems on the use of BTs in multi-robot scenarios include but are not limited to task allocation and robot coordination. In this work, we propose a market-based behavior tree method for task perception, allocation, and execution. This method introduces an auction mechanism to adjust commitments between different robots as the change of environment and extends BTs into multi-robot systems. We validate the proposed method quantitatively and qualitatively by simulation with Gazebo.
{"title":"Extending Behavior Trees with Market-Based Task Allocation in Dynamic Environments","authors":"Tao Wang, Dian-xi Shi, Wei Yi","doi":"10.1145/3440084.3441211","DOIUrl":"https://doi.org/10.1145/3440084.3441211","url":null,"abstract":"Behavior tree (BT) is a novel control architecture in the robotic field. Being modular and reactive, BTs show great advantages in autonomous systems when applied to robot control. However, there are unsolved problems on the use of BTs in multi-robot scenarios include but are not limited to task allocation and robot coordination. In this work, we propose a market-based behavior tree method for task perception, allocation, and execution. This method introduces an auction mechanism to adjust commitments between different robots as the change of environment and extends BTs into multi-robot systems. We validate the proposed method quantitatively and qualitatively by simulation with Gazebo.","PeriodicalId":250100,"journal":{"name":"Proceedings of the 2020 4th International Symposium on Computer Science and Intelligent Control","volume":"42 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-11-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125768753","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}
This paper introduces a heuristic tentacle algorithm for local path planning of unmanned skid-steering vehicle. Mobility, safety and economy are three mainly focused aspects in the navigation of unmanned ground vehicles. Critical skidding and slipping often occur during the turning motion, which effect the vehicle's motion apparently. So vehicle kinematics are discussed and applied to construct the cluster of tentacles. Several path assessment criteria named obstacle avoidance, terrain roughness and distance to the global path are discussed. Based on the multi-density clustering processed in the global path planning, heuristic method is introduced to guide to search in sparse region. The simulation analysis shows the generated local path can avoid the obstacles along the global path. Simultaneously. the global path can be smoothed through kinematic aware tentacle algorithm.
{"title":"Heuristic Tentacle Algorithm for Local Path Planning Based on Obstacles Clustering Concept","authors":"Fangxu Liu, Weiming Li, Xueyuan Li, Tianyi Bai","doi":"10.1145/3440084.3441189","DOIUrl":"https://doi.org/10.1145/3440084.3441189","url":null,"abstract":"This paper introduces a heuristic tentacle algorithm for local path planning of unmanned skid-steering vehicle. Mobility, safety and economy are three mainly focused aspects in the navigation of unmanned ground vehicles. Critical skidding and slipping often occur during the turning motion, which effect the vehicle's motion apparently. So vehicle kinematics are discussed and applied to construct the cluster of tentacles. Several path assessment criteria named obstacle avoidance, terrain roughness and distance to the global path are discussed. Based on the multi-density clustering processed in the global path planning, heuristic method is introduced to guide to search in sparse region. The simulation analysis shows the generated local path can avoid the obstacles along the global path. Simultaneously. the global path can be smoothed through kinematic aware tentacle algorithm.","PeriodicalId":250100,"journal":{"name":"Proceedings of the 2020 4th International Symposium on Computer Science and Intelligent Control","volume":"42 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-11-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115574325","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}
Irvin Dongo, Regina P. Ticona-Herrera, Yudith Cadinale, Renato Guzman
The eXtensible Markup Language (XML) has become the main standard for Web information representation and data exchange over the last decades. However, XML documents present high heterogeneity regarding their structure. Hence, there is still a need of new approaches to manage and recognize similar information that consider the content and the semantic, besides the document structure. Most current approaches semantically analyze the XML document content, regardless its structure or vice versa. In this paper, we propose LSI*, a new approach for XML documents similarity by integrating in the semantic analysis their structural composition. We extend the Latent Semantic Indexing (LSI), which is based on Singular Value Decomposition (SVD), by considering the term itself and the context (i.e., structural path) in which it appears, to determine the semantic similarity between XML documents. To evaluate the performance of our proposal, we perform experiments to compare LSI* to state-of-the-art methods based on structural and content-structural analysis. Results show a precision up to 71, 43% when the XML structure is considered in the content analysis.
{"title":"Semantic Similarity of XML Documents Based on Structural and Content Analysis","authors":"Irvin Dongo, Regina P. Ticona-Herrera, Yudith Cadinale, Renato Guzman","doi":"10.1145/3440084.3441185","DOIUrl":"https://doi.org/10.1145/3440084.3441185","url":null,"abstract":"The eXtensible Markup Language (XML) has become the main standard for Web information representation and data exchange over the last decades. However, XML documents present high heterogeneity regarding their structure. Hence, there is still a need of new approaches to manage and recognize similar information that consider the content and the semantic, besides the document structure. Most current approaches semantically analyze the XML document content, regardless its structure or vice versa. In this paper, we propose LSI*, a new approach for XML documents similarity by integrating in the semantic analysis their structural composition. We extend the Latent Semantic Indexing (LSI), which is based on Singular Value Decomposition (SVD), by considering the term itself and the context (i.e., structural path) in which it appears, to determine the semantic similarity between XML documents. To evaluate the performance of our proposal, we perform experiments to compare LSI* to state-of-the-art methods based on structural and content-structural analysis. Results show a precision up to 71, 43% when the XML structure is considered in the content analysis.","PeriodicalId":250100,"journal":{"name":"Proceedings of the 2020 4th International Symposium on Computer Science and Intelligent Control","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-11-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129401381","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}
Path planning research takes a significant position in the field of driverless driving, especially local path planning is a key point to ensure the safety of driverless vehicles. At present, the local path planning algorithm is mainly designed for the ackermann vehicle. However, due to the different steering principles, the local path planning algorithm based on the Ackerman steering principle is not suitable for the skid-steered wheeled vehicle. The local path planning algorithm does not consider the kinematics of the skid-steered wheeled vehicle. In this paper, the skid-steered wheeled vehicle is simplified to a single-axle model and the steering kinematics characteristic is analyzed. The classical Dynamic Window Approach (DWA) is improved based on the analysis of the kinematics characteristic of the skid-steered wheeled vehicle with the shortest passing time as the goal. The improved algorithm combines the available velocities of the right and left wheels at the next moment, and finally chooses the combination that takes the shortest time to avoid the obstacle. The planned local path can meet the steering requirements of the skid-steered wheeled vehicle. The improved algorithm is simulated in MATLAB, and the time passed in the map of 10m range is 42.1s.
{"title":"An Improved Local Path Planning Algorithm Based on Kinematics Analysis of The Skid-Steered Wheeled Vehicle","authors":"B. Wang, Xueyuan Li, Shihua Yuan","doi":"10.1145/3440084.3441196","DOIUrl":"https://doi.org/10.1145/3440084.3441196","url":null,"abstract":"Path planning research takes a significant position in the field of driverless driving, especially local path planning is a key point to ensure the safety of driverless vehicles. At present, the local path planning algorithm is mainly designed for the ackermann vehicle. However, due to the different steering principles, the local path planning algorithm based on the Ackerman steering principle is not suitable for the skid-steered wheeled vehicle. The local path planning algorithm does not consider the kinematics of the skid-steered wheeled vehicle. In this paper, the skid-steered wheeled vehicle is simplified to a single-axle model and the steering kinematics characteristic is analyzed. The classical Dynamic Window Approach (DWA) is improved based on the analysis of the kinematics characteristic of the skid-steered wheeled vehicle with the shortest passing time as the goal. The improved algorithm combines the available velocities of the right and left wheels at the next moment, and finally chooses the combination that takes the shortest time to avoid the obstacle. The planned local path can meet the steering requirements of the skid-steered wheeled vehicle. The improved algorithm is simulated in MATLAB, and the time passed in the map of 10m range is 42.1s.","PeriodicalId":250100,"journal":{"name":"Proceedings of the 2020 4th International Symposium on Computer Science and Intelligent Control","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-11-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129279483","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}
Jorge Ortega-Moody, Yosselin Castro Islas, Levi Howell, K. Jenab, Victoria Russ
In the food industry, what makes the difference between a company and its competitors is the quality of its food product, and one of the components that makes quality is the flavor of a product. As trends shift to meet consumer's health and wellness desires formulas are changed with companies trying to retain the same flavoring. This shifting has led to the increased demand of sensory analysis tests. Some of the limitations for sensory analysis is the required space to have the individual booths, time consuming preparation, and material costs. But even with the previous limitations, one of the most important is the training of new users and calibration of existing users. With the development of virtual reality this problem can be more easily rectified by creating a virtual scenario that utilizes all senses and reduces cost of training. The main objective of this research is the development of a virtual scenario for sensory training. This is achieved by recreating a traditional testing environment and training program to analyze scents provided by a scent generator prototype. The methodology will include the design of the training, recreating the environment, programming the interaction with the user and finally the development of a scent generator to release scents.
{"title":"Design of A Virtual Reality Scenario and Scent Generator for Sensory Training","authors":"Jorge Ortega-Moody, Yosselin Castro Islas, Levi Howell, K. Jenab, Victoria Russ","doi":"10.1145/3440084.3441218","DOIUrl":"https://doi.org/10.1145/3440084.3441218","url":null,"abstract":"In the food industry, what makes the difference between a company and its competitors is the quality of its food product, and one of the components that makes quality is the flavor of a product. As trends shift to meet consumer's health and wellness desires formulas are changed with companies trying to retain the same flavoring. This shifting has led to the increased demand of sensory analysis tests. Some of the limitations for sensory analysis is the required space to have the individual booths, time consuming preparation, and material costs. But even with the previous limitations, one of the most important is the training of new users and calibration of existing users. With the development of virtual reality this problem can be more easily rectified by creating a virtual scenario that utilizes all senses and reduces cost of training. The main objective of this research is the development of a virtual scenario for sensory training. This is achieved by recreating a traditional testing environment and training program to analyze scents provided by a scent generator prototype. The methodology will include the design of the training, recreating the environment, programming the interaction with the user and finally the development of a scent generator to release scents.","PeriodicalId":250100,"journal":{"name":"Proceedings of the 2020 4th International Symposium on Computer Science and Intelligent Control","volume":"96 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-11-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123339697","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}
T. Krajník, Tomáš Vintr, G. Broughton, Filip Majer, Tomáš Rouček, Jirí Ulrich, Jan Blaha, Veronika Pěčonková, Martin Rektoris
Chronorobotics is the investigation of scientific methods allowing robots to adapt to and learn from the perpetual changes occurring in natural and human-populated environments. We present methods that can introduce the notion of dynamics into spatial environment models, resulting in representations which provide service robots with the ability to predict future states of changing environments. Several long-term experiments indicate that the aforementioned methods gradually improve the efficiency of robots' autonomous operations over time. More importantly, the experiments indicate that chronorobotic concepts improve robots' ability to seamlessly merge into human-populated environments, which is important for their integration and acceptance in human societies.
{"title":"CHRONOROBOTICS","authors":"T. Krajník, Tomáš Vintr, G. Broughton, Filip Majer, Tomáš Rouček, Jirí Ulrich, Jan Blaha, Veronika Pěčonková, Martin Rektoris","doi":"10.1145/3440084.3441195","DOIUrl":"https://doi.org/10.1145/3440084.3441195","url":null,"abstract":"Chronorobotics is the investigation of scientific methods allowing robots to adapt to and learn from the perpetual changes occurring in natural and human-populated environments. We present methods that can introduce the notion of dynamics into spatial environment models, resulting in representations which provide service robots with the ability to predict future states of changing environments. Several long-term experiments indicate that the aforementioned methods gradually improve the efficiency of robots' autonomous operations over time. More importantly, the experiments indicate that chronorobotic concepts improve robots' ability to seamlessly merge into human-populated environments, which is important for their integration and acceptance in human societies.","PeriodicalId":250100,"journal":{"name":"Proceedings of the 2020 4th International Symposium on Computer Science and Intelligent Control","volume":"45 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-11-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114191789","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}
In the Natural Language Processing (NLP) community, automatic text summarization is considered to be a very difficult problem. The textual content on the web, in particular, is growing at an exponential rate. The ability to decipher through such massive amount of data, in order to extract the useful information, is a major undertaking and requires an automatic mechanism to aid with the extant repository of information. A good text summarization system should understand the whole text, reorganize information, and generate coherent, informative and remarkably short summaries to convey the important information of the original text. In this paper, an innovative text summarization model has been constructed, which combines BERT, reinforcement learning, sequence-to-sequence and other technologies. Our model is evaluated on the LCSTS[1] dataset, which is a high-quality corpus of Chinese short text summarization dataset constructed from "Sina Weibo", The experiment shows that our method has made a great breakthrough in Rouge Scores compared with other researches.
{"title":"Automatic Text Summarization on Social Media","authors":"Zhang Kerui, H. Haichao, Li Yuxia","doi":"10.1145/3440084.3441182","DOIUrl":"https://doi.org/10.1145/3440084.3441182","url":null,"abstract":"In the Natural Language Processing (NLP) community, automatic text summarization is considered to be a very difficult problem. The textual content on the web, in particular, is growing at an exponential rate. The ability to decipher through such massive amount of data, in order to extract the useful information, is a major undertaking and requires an automatic mechanism to aid with the extant repository of information. A good text summarization system should understand the whole text, reorganize information, and generate coherent, informative and remarkably short summaries to convey the important information of the original text. In this paper, an innovative text summarization model has been constructed, which combines BERT, reinforcement learning, sequence-to-sequence and other technologies. Our model is evaluated on the LCSTS[1] dataset, which is a high-quality corpus of Chinese short text summarization dataset constructed from \"Sina Weibo\", The experiment shows that our method has made a great breakthrough in Rouge Scores compared with other researches.","PeriodicalId":250100,"journal":{"name":"Proceedings of the 2020 4th International Symposium on Computer Science and Intelligent Control","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-11-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129256140","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}
Aiming at the problems of electronic equipment long service life, less failure data and the reliability prediction, based on the analysis of physics of failure mechanism, a correlation competing failure model for the performance degradation process of electronic equipment under multi-type shocks is established, and the dynamic threshold is analyzed to obtain a more realistic reliability model. On the basis of stress damage analysis and structural analysis, this article presents the pre-evaluation process, and applies a kind of improved entropy weight method to comprehensively select and calculate the parameters describing the overall performance. Finally, by simulating the degradation process of the laser, the variation rule of its reliability is obtained, and the rationality and validity of the model are verified.
{"title":"Research on Performance Degradation of Electronic Equipment Under Multi-type Shocks","authors":"Dongdong Zhang, Xiaochuan Ai","doi":"10.1145/3440084.3441208","DOIUrl":"https://doi.org/10.1145/3440084.3441208","url":null,"abstract":"Aiming at the problems of electronic equipment long service life, less failure data and the reliability prediction, based on the analysis of physics of failure mechanism, a correlation competing failure model for the performance degradation process of electronic equipment under multi-type shocks is established, and the dynamic threshold is analyzed to obtain a more realistic reliability model. On the basis of stress damage analysis and structural analysis, this article presents the pre-evaluation process, and applies a kind of improved entropy weight method to comprehensively select and calculate the parameters describing the overall performance. Finally, by simulating the degradation process of the laser, the variation rule of its reliability is obtained, and the rationality and validity of the model are verified.","PeriodicalId":250100,"journal":{"name":"Proceedings of the 2020 4th International Symposium on Computer Science and Intelligent Control","volume":"47 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-11-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134160750","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}
Object detection, as an important branch of computer vision, has been widely studied in recent years. However, the lack of large labeled dataset obstructs the usage of convolutional neural networks (CNN) for detecting in thermal infrared (TIR) images. Most existing dataset focus on visible images, while thermal infrared images are helpful for detection even in a dark environment. To address this problem, we propose to use image-to-image translation models. These models allow us to translate the available labeled visible images to synthetic infrared images. Based on the original pedestrian dataset CVC-09, we use the pedestrian dataset CVC-14 to generate some labeled pedestrian infrared images. Finally, we compare original dataset with classic data augmentation and synthetic data augmentation training CNN. In addition, we explore the quality of synthetic TIR images using contrast experiments. The average precision of detection using classic data augmentation alone is 79.18%. By adding synthetic data augmentation, the average precision has improved to 82.24%. We believe that this method of synthetic data augmentation can be extended to other infrared detection applications and achieve other breakthroughs.
{"title":"Infrared Pedestrian Detection Based on GAN Data Augmentation","authors":"Jinda Hu, Yanshun Zhao, Xindong Zhang","doi":"10.1145/3440084.3441178","DOIUrl":"https://doi.org/10.1145/3440084.3441178","url":null,"abstract":"Object detection, as an important branch of computer vision, has been widely studied in recent years. However, the lack of large labeled dataset obstructs the usage of convolutional neural networks (CNN) for detecting in thermal infrared (TIR) images. Most existing dataset focus on visible images, while thermal infrared images are helpful for detection even in a dark environment. To address this problem, we propose to use image-to-image translation models. These models allow us to translate the available labeled visible images to synthetic infrared images. Based on the original pedestrian dataset CVC-09, we use the pedestrian dataset CVC-14 to generate some labeled pedestrian infrared images. Finally, we compare original dataset with classic data augmentation and synthetic data augmentation training CNN. In addition, we explore the quality of synthetic TIR images using contrast experiments. The average precision of detection using classic data augmentation alone is 79.18%. By adding synthetic data augmentation, the average precision has improved to 82.24%. We believe that this method of synthetic data augmentation can be extended to other infrared detection applications and achieve other breakthroughs.","PeriodicalId":250100,"journal":{"name":"Proceedings of the 2020 4th International Symposium on Computer Science and Intelligent Control","volume":"412 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-11-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"120952390","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：