Atherosclerosis is a disease in which plaque builds up inside the arteries that eventually intend to be hardened over time thus narrows the arteries. This work is to develop an atherosclerosis diagnosis system using fuzzy logic and to construct a prototype of atherosclerosis risk monitor. Simultaneously, the system is validated by comparing the result obtained from the system and the one from the prototype. Five membership functions are setup such as age, body mass index (BMI), intima-media thickness (IMT), blood pressure and blood glucose level, from 20 volunteers males and females. The output fuzzy sets are divided into Healthy, Low risk, Mid risk, High risk, and Very high risk. This leads to the awareness level of people thus improve the societal health. The constructed prototype is validated by inserting the data and reconfirmed the diagnosed result on the display. It showed that the prototype has 95% realibility since the output of the system is calculated with 100% low risk (LR) of atherosclerosis.
{"title":"Initial Study to Evaluate Fuzzy Logic on Diagnosis of Generic Atherosclerosis","authors":"Nabilah Ibrahim, Lina Farhana Mahadi, F. Mahmud","doi":"10.1109/ICVISP.2017.31","DOIUrl":"https://doi.org/10.1109/ICVISP.2017.31","url":null,"abstract":"Atherosclerosis is a disease in which plaque builds up inside the arteries that eventually intend to be hardened over time thus narrows the arteries. This work is to develop an atherosclerosis diagnosis system using fuzzy logic and to construct a prototype of atherosclerosis risk monitor. Simultaneously, the system is validated by comparing the result obtained from the system and the one from the prototype. Five membership functions are setup such as age, body mass index (BMI), intima-media thickness (IMT), blood pressure and blood glucose level, from 20 volunteers males and females. The output fuzzy sets are divided into Healthy, Low risk, Mid risk, High risk, and Very high risk. This leads to the awareness level of people thus improve the societal health. The constructed prototype is validated by inserting the data and reconfirmed the diagnosed result on the display. It showed that the prototype has 95% realibility since the output of the system is calculated with 100% low risk (LR) of atherosclerosis.","PeriodicalId":404467,"journal":{"name":"2017 International Conference on Vision, Image and Signal Processing (ICVISP)","volume":"68 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117144710","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}
For the operation and aerial photography of the UAV, it is important to identify the blindspots and observe the details on the ground. But limited by the camera resolution, small or fuzzy objects can not be effectively observed. Therefore, repairment of high-definition images has become one of the important problems to be solved. In recent years, the development of the deep learning method has effectively solved the loss and blurring of images, but because of the difficulties in training and the speed of calculation it can only be used with low-pixel images. Therefore, we propose a method for superimposing images first with the content and textual recovery for the defaced area. We use unsupervised learning GANs and trained VGG network to restore holes and missing areas of the image, and then enlarge it through CNN method. Our preliminary results show that high resolution image restoration speed has been greatly improved, and details become sharper than using traditional method.
{"title":"High-Resolution Image Inpainting through Multiple Deep Networks","authors":"Chih-Hsu Hsu, Feng Chen, Guijin Wang","doi":"10.1109/ICVISP.2017.27","DOIUrl":"https://doi.org/10.1109/ICVISP.2017.27","url":null,"abstract":"For the operation and aerial photography of the UAV, it is important to identify the blindspots and observe the details on the ground. But limited by the camera resolution, small or fuzzy objects can not be effectively observed. Therefore, repairment of high-definition images has become one of the important problems to be solved. In recent years, the development of the deep learning method has effectively solved the loss and blurring of images, but because of the difficulties in training and the speed of calculation it can only be used with low-pixel images. Therefore, we propose a method for superimposing images first with the content and textual recovery for the defaced area. We use unsupervised learning GANs and trained VGG network to restore holes and missing areas of the image, and then enlarge it through CNN method. Our preliminary results show that high resolution image restoration speed has been greatly improved, and details become sharper than using traditional method.","PeriodicalId":404467,"journal":{"name":"2017 International Conference on Vision, Image and Signal Processing (ICVISP)","volume":"13 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115072568","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}
Ramanpreet Singh, Michael Thompson, S. A. Mathews, O. Agbogidi, Kunal Bhadane, K. Namuduri
During disasters such as hurricanes, earthquakes, tornadoes, storms, tsunamis and man-made disasters, the need for communication becomes more imperative for the first responders as well as the people who were affected. It is very common that during such situations communication is interrupted due to physical damage of network devices, network congestion and interruption of supporting network infrastructure. Communications during such incidents can be the distinguishing factor between life and death for the affected population. Without communications, first responders might find it difficult to locate and rescue trapped victims during disaster recovery operations due to lack of synchronized operations. This paper presents a summary of our research efforts that led to the design of an integrated aerial platform for enabling communications and information sharing among the first responders and citizens during disaster recovery operations.
{"title":"Aerial Base Stations for Enabling Cellular Communications during Emergency Situation","authors":"Ramanpreet Singh, Michael Thompson, S. A. Mathews, O. Agbogidi, Kunal Bhadane, K. Namuduri","doi":"10.1109/ICVISP.2017.24","DOIUrl":"https://doi.org/10.1109/ICVISP.2017.24","url":null,"abstract":"During disasters such as hurricanes, earthquakes, tornadoes, storms, tsunamis and man-made disasters, the need for communication becomes more imperative for the first responders as well as the people who were affected. It is very common that during such situations communication is interrupted due to physical damage of network devices, network congestion and interruption of supporting network infrastructure. Communications during such incidents can be the distinguishing factor between life and death for the affected population. Without communications, first responders might find it difficult to locate and rescue trapped victims during disaster recovery operations due to lack of synchronized operations. This paper presents a summary of our research efforts that led to the design of an integrated aerial platform for enabling communications and information sharing among the first responders and citizens during disaster recovery operations.","PeriodicalId":404467,"journal":{"name":"2017 International Conference on Vision, Image and Signal Processing (ICVISP)","volume":"107 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116374179","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}
Garvit Joshi, Chaitanya Dandvate, H. Tiwari, Aakash Mundhare
Sometimes parents don't have resources or time to attend to their young ones as they have certain predispositions. This document demonstrates the process of construction of a web-service/module, defines the algorithm, procedure of construction of the algorithm and the analysis/results of the procedures performed. The market for this system is the working class nuclear families or single parents that are not present for their babies and have to take help from nannies to keep an eye for them. The algorithm constructed is itself build upon various algorithms that were developed in past and incorporated in the ML studio as modules so a dataset has been generated and utilized these modules to from an algorithm to predict the probability of a child's crying in next few hours based on the previous data that has been collected (randomly generated in this case). A module for creation of automatic machine is stated which augment a basic child cry monitor with Machine Learning and Cognitive services for faster cheaper and more reliable cloud based solution for parents.
{"title":"Prediction of Probability of Crying of a Child and System Formation for Cry Detection and Financial Viability of the System","authors":"Garvit Joshi, Chaitanya Dandvate, H. Tiwari, Aakash Mundhare","doi":"10.1109/ICVISP.2017.33","DOIUrl":"https://doi.org/10.1109/ICVISP.2017.33","url":null,"abstract":"Sometimes parents don't have resources or time to attend to their young ones as they have certain predispositions. This document demonstrates the process of construction of a web-service/module, defines the algorithm, procedure of construction of the algorithm and the analysis/results of the procedures performed. The market for this system is the working class nuclear families or single parents that are not present for their babies and have to take help from nannies to keep an eye for them. The algorithm constructed is itself build upon various algorithms that were developed in past and incorporated in the ML studio as modules so a dataset has been generated and utilized these modules to from an algorithm to predict the probability of a child's crying in next few hours based on the previous data that has been collected (randomly generated in this case). A module for creation of automatic machine is stated which augment a basic child cry monitor with Machine Learning and Cognitive services for faster cheaper and more reliable cloud based solution for parents.","PeriodicalId":404467,"journal":{"name":"2017 International Conference on Vision, Image and Signal Processing (ICVISP)","volume":"22 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133931414","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}
M. Saealal, Dafizal Derawi, Nurul Dayana Salim, M. Tumari
This paper presents the real-time implementation of a powerful nonlinear complementary filter on special orthogonal group of rotation matrices, called as NCF SO(3) for attitude estimation. It fuses the raw data from accelerometers, magnetometer, and gyroscopes sensors to get reliable real-time attitude estimation. Gyroscopes is used as the main sensor for attitude estimation and another two sensors are used to correct drift error of gyroscopes. In this paper, the performance of NCF SO(3) is explored on performance in highly dynamic manoeuvres in real-time. Real-time experiments were conducted to compare its performance with conventional Extended Kalman Filter (EKF) to exploit the positive features of NCF SO(3) for small-scale aerial robot with limited on-board processor memory cases. The experimental results show the proposed real-time filter has excellent estimated attitude data and can reduce the computational cost, compared to EKF. Thus, it is suitable for small-scale aerial robot which has memory limitation of on-board processor.
{"title":"Real-Time Nonlinear Complementary Filter on SO(3) for Attitude Estimation of Small-Scale Aerial Robot","authors":"M. Saealal, Dafizal Derawi, Nurul Dayana Salim, M. Tumari","doi":"10.1109/ICVISP.2017.25","DOIUrl":"https://doi.org/10.1109/ICVISP.2017.25","url":null,"abstract":"This paper presents the real-time implementation of a powerful nonlinear complementary filter on special orthogonal group of rotation matrices, called as NCF SO(3) for attitude estimation. It fuses the raw data from accelerometers, magnetometer, and gyroscopes sensors to get reliable real-time attitude estimation. Gyroscopes is used as the main sensor for attitude estimation and another two sensors are used to correct drift error of gyroscopes. In this paper, the performance of NCF SO(3) is explored on performance in highly dynamic manoeuvres in real-time. Real-time experiments were conducted to compare its performance with conventional Extended Kalman Filter (EKF) to exploit the positive features of NCF SO(3) for small-scale aerial robot with limited on-board processor memory cases. The experimental results show the proposed real-time filter has excellent estimated attitude data and can reduce the computational cost, compared to EKF. Thus, it is suitable for small-scale aerial robot which has memory limitation of on-board processor.","PeriodicalId":404467,"journal":{"name":"2017 International Conference on Vision, Image and Signal Processing (ICVISP)","volume":"68 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125215193","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}
Traffic sign detection and recognition play crucial roles on the Intelligent Transportation System(ITS). Firstly, in YCbCr color space, color segmentation of the traffic scene images using Gaussian color model is calculated Cand traffic sign regions are obtained. Secondly, the morphology processing is utilized on the segmented image to extract the candidate traffic signs with a rectangle region in the original image to be selected according as its shape property. Finally, template matching is applied for speed signs classification. The performance of the proposed method is evaluated on Norwegian speed limit signs in natural environment. Experiment results show that this algorithm can effectively improve the traffic sign detection efficiency, which is used in traffic signs recognition and tracking of intelligent vehicles.
{"title":"Speed Limit Sign Detection Based on Gaussian Color Model and Template Matching","authors":"Han Huang, Ling-Ying Hou","doi":"10.1109/ICVISP.2017.30","DOIUrl":"https://doi.org/10.1109/ICVISP.2017.30","url":null,"abstract":"Traffic sign detection and recognition play crucial roles on the Intelligent Transportation System(ITS). Firstly, in YCbCr color space, color segmentation of the traffic scene images using Gaussian color model is calculated Cand traffic sign regions are obtained. Secondly, the morphology processing is utilized on the segmented image to extract the candidate traffic signs with a rectangle region in the original image to be selected according as its shape property. Finally, template matching is applied for speed signs classification. The performance of the proposed method is evaluated on Norwegian speed limit signs in natural environment. Experiment results show that this algorithm can effectively improve the traffic sign detection efficiency, which is used in traffic signs recognition and tracking of intelligent vehicles.","PeriodicalId":404467,"journal":{"name":"2017 International Conference on Vision, Image and Signal Processing (ICVISP)","volume":"23 15","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"120851494","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}
Shilong Liu, Hongkun Wu, Ruowei Li, Md. Arifur Rahman, Xuan He, S. Liu, N. Kwok
Digital images captured in hazy conditions suffer from colour distortion and loss of contrast, posing difficulties in being applied for further applications. Due to the existed challenge and its great significance, a large amount of research has been conducted for image de-hazing. Among the image haze removal methods, the algorithm based on dark channel prior is proved to be the most effective. Furthermore, the introduction of guided filter has boosted its efficiency to a large extent. However, the requirement for transmission refinement and the assumption that the transmission is the same in each colour channel still make the DCP concept based methods time consuming and suffer from colour distortion. To solve this problem, an approach named as Image De-hazing Based on Polynomial Estimation and Steepest Descent Concept (IDBPESDC) is proposed, which derives the pixel-wised transmission that does not require any further refinement. Additionally, image de-hazing procedures based on the steepest descent concept are adopted so that the objective of saturation enhancement under the minimum hue change constraint is achieved. Experiments are conducted on one hundred hazy images, processed by the proposed method and four other available approaches. Results are analysed qualitatively and quantitatively, which verified the effectiveness and efficiency of the proposed algorithm.
{"title":"Image De-hazing Based on Polynomial Estimation and Steepest Descent Concept","authors":"Shilong Liu, Hongkun Wu, Ruowei Li, Md. Arifur Rahman, Xuan He, S. Liu, N. Kwok","doi":"10.1109/ICVISP.2017.17","DOIUrl":"https://doi.org/10.1109/ICVISP.2017.17","url":null,"abstract":"Digital images captured in hazy conditions suffer from colour distortion and loss of contrast, posing difficulties in being applied for further applications. Due to the existed challenge and its great significance, a large amount of research has been conducted for image de-hazing. Among the image haze removal methods, the algorithm based on dark channel prior is proved to be the most effective. Furthermore, the introduction of guided filter has boosted its efficiency to a large extent. However, the requirement for transmission refinement and the assumption that the transmission is the same in each colour channel still make the DCP concept based methods time consuming and suffer from colour distortion. To solve this problem, an approach named as Image De-hazing Based on Polynomial Estimation and Steepest Descent Concept (IDBPESDC) is proposed, which derives the pixel-wised transmission that does not require any further refinement. Additionally, image de-hazing procedures based on the steepest descent concept are adopted so that the objective of saturation enhancement under the minimum hue change constraint is achieved. Experiments are conducted on one hundred hazy images, processed by the proposed method and four other available approaches. Results are analysed qualitatively and quantitatively, which verified the effectiveness and efficiency of the proposed algorithm.","PeriodicalId":404467,"journal":{"name":"2017 International Conference on Vision, Image and Signal Processing (ICVISP)","volume":"17 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124939102","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}
Cunguang Zhang, Bo Li, Hongxu Jiang, Huiyong Li, Jiao Chen
Sea land segmentation is an essential part in remote sensing image processing, greatly reducing the interesting area. However, faced with the faster real-time processing of remote sensing images, the SLS algorithm's throughputs is always constrained by the self complexity and the on-chip resources. A new method based on MIMO for balancing resource and throughputs is proposed to segment the sea and land via edge detection. Firstly, according to the situation that the input and output data rate is not equal, the principle of MIMO is introduced to redesign the edge detection structure to double the bandwidth. Second, multi-port cache brings storage resources, which have to be reused, and reduce resource usage. Finally, the parallel sliding widdows method is adopted to usd in piplines of image outputing. The experimental results show that, compared with the state-of-art structure, this method performed better in hardware utilization and bandwidth.
{"title":"High Throughput Hardware Architecture of a MIMO-based Sea Land Segmentation for On-Orbit Remote Sensing Image Processing","authors":"Cunguang Zhang, Bo Li, Hongxu Jiang, Huiyong Li, Jiao Chen","doi":"10.1109/ICVISP.2017.21","DOIUrl":"https://doi.org/10.1109/ICVISP.2017.21","url":null,"abstract":"Sea land segmentation is an essential part in remote sensing image processing, greatly reducing the interesting area. However, faced with the faster real-time processing of remote sensing images, the SLS algorithm's throughputs is always constrained by the self complexity and the on-chip resources. A new method based on MIMO for balancing resource and throughputs is proposed to segment the sea and land via edge detection. Firstly, according to the situation that the input and output data rate is not equal, the principle of MIMO is introduced to redesign the edge detection structure to double the bandwidth. Second, multi-port cache brings storage resources, which have to be reused, and reduce resource usage. Finally, the parallel sliding widdows method is adopted to usd in piplines of image outputing. The experimental results show that, compared with the state-of-art structure, this method performed better in hardware utilization and bandwidth.","PeriodicalId":404467,"journal":{"name":"2017 International Conference on Vision, Image and Signal Processing (ICVISP)","volume":"2 5 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126098822","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 presents the utilization of an inertial measurement unit (IMU) for wind-characterization applications. A method of wind velocity estimation using an IMU installed on a tethered balloon is described. The system of a balloon tethered to a ground station acts as a measuring instrument for wind approximation. The wind-responding motion of tethered balloon is captured with the utilization of a standard set of on-board sensors (i.e., accelerometers, gyroscope, thermometer, and barometer). An Extended Kalman Filter (EKF) technique is implemented for the wind estimation with the use of an estimated motion of spherical balloon. The performance of the developed method in order to measure the wind speed in variations of frequencies is determined. The limitation of the system performance is found to be due to the delay in the filter and performance of the selected sensor set.
{"title":"A Method of Wind Velocity Estimation Using a Tethered, Spherical Balloon with Standard Aviation Sensors","authors":"T. Yomchinda","doi":"10.1109/ICVISP.2017.32","DOIUrl":"https://doi.org/10.1109/ICVISP.2017.32","url":null,"abstract":"This paper presents the utilization of an inertial measurement unit (IMU) for wind-characterization applications. A method of wind velocity estimation using an IMU installed on a tethered balloon is described. The system of a balloon tethered to a ground station acts as a measuring instrument for wind approximation. The wind-responding motion of tethered balloon is captured with the utilization of a standard set of on-board sensors (i.e., accelerometers, gyroscope, thermometer, and barometer). An Extended Kalman Filter (EKF) technique is implemented for the wind estimation with the use of an estimated motion of spherical balloon. The performance of the developed method in order to measure the wind speed in variations of frequencies is determined. The limitation of the system performance is found to be due to the delay in the filter and performance of the selected sensor set.","PeriodicalId":404467,"journal":{"name":"2017 International Conference on Vision, Image and Signal Processing (ICVISP)","volume":"2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126237833","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}
Jiajun Wang, Beizhan Wang, Yinhuan Zheng, Weiqiang Liu
At present, the combination of deep learning and traditional method are used to solve the problem of face detection that caused by image quality, various poses, occlusions, complex facial expressions, illumination and real time detect. In this paper, we use a face detection framework based on cascaded convolutional neural network, which is used to balance the accuracy and running time cost. We use full convolutional neural network (FCN) to extract candidate regions of human face in the first stage, which is more efficient than selective search, Edge Box and other algorithms. Combining with the NMS algorithm and bounding box regression during the whole process, we can get more accurate face position. In order to improve the accuracy and enhance the ability of the algorithm to distinguish the face, we improved the training method, optimized the training set, and used the multi-task learning network. Experiment results show that the framework has higher accuracy and costs shorter time in face detection on FDDB.
{"title":"Research and Implementation on Face Detection Approach Based on Cascaded Convolutional Neural Networks","authors":"Jiajun Wang, Beizhan Wang, Yinhuan Zheng, Weiqiang Liu","doi":"10.1109/ICVISP.2017.10","DOIUrl":"https://doi.org/10.1109/ICVISP.2017.10","url":null,"abstract":"At present, the combination of deep learning and traditional method are used to solve the problem of face detection that caused by image quality, various poses, occlusions, complex facial expressions, illumination and real time detect. In this paper, we use a face detection framework based on cascaded convolutional neural network, which is used to balance the accuracy and running time cost. We use full convolutional neural network (FCN) to extract candidate regions of human face in the first stage, which is more efficient than selective search, Edge Box and other algorithms. Combining with the NMS algorithm and bounding box regression during the whole process, we can get more accurate face position. In order to improve the accuracy and enhance the ability of the algorithm to distinguish the face, we improved the training method, optimized the training set, and used the multi-task learning network. Experiment results show that the framework has higher accuracy and costs shorter time in face detection on FDDB.","PeriodicalId":404467,"journal":{"name":"2017 International Conference on Vision, Image and Signal Processing (ICVISP)","volume":"23 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130013450","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：