Pub Date : 2023-03-01DOI: 10.1109/PRMVIA58252.2023.00036
Chuancheng Deng, Feng Yi, Xing Li, Junyi Tang, Guang Sun
Decision tree algorithm belongs to the machine learning method of supervised learning, and it is one of the common techniques of data mining. CHAID algorithm is one of the common decision tree algorithms. The CHAID algorithm can be used for prediction as well as for classification. It is mainly used in market analysis, risk prediction and other fields. The second part tells the core idea of CHAID algorithm and the specific steps of the classification process, as well as the principle formula of chi-square detection and the specific steps of the calculation mainly used by CHAID algorithm. The third part describes the branch principle of CHAID algorithm and other decision tree algorithm, and analyses the accuracy of CHAID algorithm. Let us choose a relatively good decision tree algorithm according to the specific data situation, and apply the CHAID algorithm can also make some countermeasures according to some factors affecting the accuracy, which is more conducive to get a more accurate and better result.
{"title":"Performance Analysis of CHAID Algorithm for Accuracy","authors":"Chuancheng Deng, Feng Yi, Xing Li, Junyi Tang, Guang Sun","doi":"10.1109/PRMVIA58252.2023.00036","DOIUrl":"https://doi.org/10.1109/PRMVIA58252.2023.00036","url":null,"abstract":"Decision tree algorithm belongs to the machine learning method of supervised learning, and it is one of the common techniques of data mining. CHAID algorithm is one of the common decision tree algorithms. The CHAID algorithm can be used for prediction as well as for classification. It is mainly used in market analysis, risk prediction and other fields. The second part tells the core idea of CHAID algorithm and the specific steps of the classification process, as well as the principle formula of chi-square detection and the specific steps of the calculation mainly used by CHAID algorithm. The third part describes the branch principle of CHAID algorithm and other decision tree algorithm, and analyses the accuracy of CHAID algorithm. Let us choose a relatively good decision tree algorithm according to the specific data situation, and apply the CHAID algorithm can also make some countermeasures according to some factors affecting the accuracy, which is more conducive to get a more accurate and better result.","PeriodicalId":221346,"journal":{"name":"2023 International Conference on Pattern Recognition, Machine Vision and Intelligent Algorithms (PRMVIA)","volume":"86 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115621306","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 : 2023-03-01DOI: 10.1109/prmvia58252.2023.00019
Fengze Zhong
The family living environment today has problems in safety and convenience, and the existing smart home system generally cannot guarantee high security, low energy consumption and accurate detection. A smart home safety environment detection system based on the AT89C51 microcontroller is proposed to solve it. The design uses a ZPH01 PM2.5 detector, MQ7 carbon monoxide (CO) sensor, DHT11 temperature and humidity sensor, and MQ2 smoke sensor. It can achieve the detection, display and alert of indoor temperature, humidity, carbon monoxide concentration, PM2.5 concentration and smoke concentration. At the same time, the HC-SR501 human body sensor module is used to detect the movement of the indoor area in real-time and send alerts. Also, the Principal Component Analysis (PCA) face recognition method is used to realize the recognition of humans at access control. The simulation results show that the designed system can detect the quality of the home environment in real-time and identify the personnel, significantly improving the home environment’s safety factor and quality of life.
{"title":"Design of A Smart Home Environment Monitoring System Based on Single-chip Microcomputer","authors":"Fengze Zhong","doi":"10.1109/prmvia58252.2023.00019","DOIUrl":"https://doi.org/10.1109/prmvia58252.2023.00019","url":null,"abstract":"The family living environment today has problems in safety and convenience, and the existing smart home system generally cannot guarantee high security, low energy consumption and accurate detection. A smart home safety environment detection system based on the AT89C51 microcontroller is proposed to solve it. The design uses a ZPH01 PM2.5 detector, MQ7 carbon monoxide (CO) sensor, DHT11 temperature and humidity sensor, and MQ2 smoke sensor. It can achieve the detection, display and alert of indoor temperature, humidity, carbon monoxide concentration, PM2.5 concentration and smoke concentration. At the same time, the HC-SR501 human body sensor module is used to detect the movement of the indoor area in real-time and send alerts. Also, the Principal Component Analysis (PCA) face recognition method is used to realize the recognition of humans at access control. The simulation results show that the designed system can detect the quality of the home environment in real-time and identify the personnel, significantly improving the home environment’s safety factor and quality of life.","PeriodicalId":221346,"journal":{"name":"2023 International Conference on Pattern Recognition, Machine Vision and Intelligent Algorithms (PRMVIA)","volume":"51 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130260181","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}
The calculation method for hydrostatic buoyant center and float status of amphibious vehicle based on three-dimensional model, mass and center of mass was studied. The condition of hydrostatic equilibrium of amphibious vehicle was introduced. The calculation method of buoyant center using the CATIA program and the computing process of hydrostatic float status were researched. The secondary development for CATIA was processed based on Python. The calculation program for hydrostatic float status of amphibious vehicle was written at last. The accurate and efficient calculation for hydrostatic performance of amphibious vehicle was realized.
{"title":"Study on Calculation Method for Hydrostatic Performance of Amphibious Vehicle","authors":"Qinghui Zhang, Xinxin Liu, Xin Zhao, Hongbin Xu, Zhengyu Li, Xiaolei Li","doi":"10.1109/PRMVIA58252.2023.00027","DOIUrl":"https://doi.org/10.1109/PRMVIA58252.2023.00027","url":null,"abstract":"The calculation method for hydrostatic buoyant center and float status of amphibious vehicle based on three-dimensional model, mass and center of mass was studied. The condition of hydrostatic equilibrium of amphibious vehicle was introduced. The calculation method of buoyant center using the CATIA program and the computing process of hydrostatic float status were researched. The secondary development for CATIA was processed based on Python. The calculation program for hydrostatic float status of amphibious vehicle was written at last. The accurate and efficient calculation for hydrostatic performance of amphibious vehicle was realized.","PeriodicalId":221346,"journal":{"name":"2023 International Conference on Pattern Recognition, Machine Vision and Intelligent Algorithms (PRMVIA)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129347527","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 : 2023-03-01DOI: 10.1109/prmvia58252.2023.00012
Feiyang Qin, Wenqi Na, Song Gao, Shaowen Yao
Although deep neural networks (DNNs) have achieved exceptional performance, they are shown to be fragile to universal adversarial perturbations (UAP), which can be applied to any images to fool a well-trained DNN. Several methods have been proposed to design universal perturbations. However, these methods often leave visible traces in natural images. In this paper, we propose Sigma-UAP, a semi-universal adversarial attack, to enhance the quasi-imperceptibility of universal adversarial perturbations, in which the Sigma-map algorithm is leveraged to hide perturbations by identifying the low-frequency region of the image and eliminating the perturbations in that region. Then, we use a simple matrix calculation to augment the perturbation in the high-frequency region to ensure the attack effectiveness of the perturbation. The extensive empirical experiments show that, compared with the state-of-the-art universal adversarial attacks, Sigma-UAP method obtains excellent attack capabilities in visual effect and attack success rate.
{"title":"Sigma-UAP: An Invisible Semi-Universal Adversarial Attack Against Deep Neural Networks","authors":"Feiyang Qin, Wenqi Na, Song Gao, Shaowen Yao","doi":"10.1109/prmvia58252.2023.00012","DOIUrl":"https://doi.org/10.1109/prmvia58252.2023.00012","url":null,"abstract":"Although deep neural networks (DNNs) have achieved exceptional performance, they are shown to be fragile to universal adversarial perturbations (UAP), which can be applied to any images to fool a well-trained DNN. Several methods have been proposed to design universal perturbations. However, these methods often leave visible traces in natural images. In this paper, we propose Sigma-UAP, a semi-universal adversarial attack, to enhance the quasi-imperceptibility of universal adversarial perturbations, in which the Sigma-map algorithm is leveraged to hide perturbations by identifying the low-frequency region of the image and eliminating the perturbations in that region. Then, we use a simple matrix calculation to augment the perturbation in the high-frequency region to ensure the attack effectiveness of the perturbation. The extensive empirical experiments show that, compared with the state-of-the-art universal adversarial attacks, Sigma-UAP method obtains excellent attack capabilities in visual effect and attack success rate.","PeriodicalId":221346,"journal":{"name":"2023 International Conference on Pattern Recognition, Machine Vision and Intelligent Algorithms (PRMVIA)","volume":"63 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114068438","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 : 2023-03-01DOI: 10.1109/prmvia58252.2023.00046
Yu Zhang, Miao Dong
SLAM (simultaneous localization and mapping), will further known as synchronous localization and mapping, is a technology that is used to tackle the issue of localization and map building while a robot travels in an unfamiliar environment. Traditional SLAM relies on point features to estimate camera pose, which makes it difficult to extract enough point features in low-texture scenes. When the camera shakes violently or rotates too fast, the robustness of a point-based SLAM system is poor. Aiming at the problem of poor robustness of the existing visual SLAM (synchronous localization and mapping technology) system, based on the ORB-SLAM3 framework, the point feature extractor is replaced with a self-supervised deep neural network, and a matching filtering algorithm based on threshold and motion statistics is proposed to eliminate point mismatch, this significantly accelerates the system’s real- time and accuracy. Likewise, linear activities are integrated into the front-end information extraction, a linear feature extraction model is established, approximation linear features are merged and processed, and the linear feature description and mismatching eradication process are simplified. Finally, the weight allocation idea is introduced into the construction of the point and line error model, and the weight of the point and line is reasonably allocated according to the richness of the scene. Experiments on absolute error trajectory on the TUM dataset emphasize that the revised algorithm increased efficiency and stability when compared to the ORB-SLAM3 system.
{"title":"Research on visual SLAM algorithm based on improved point-line feature fusion","authors":"Yu Zhang, Miao Dong","doi":"10.1109/prmvia58252.2023.00046","DOIUrl":"https://doi.org/10.1109/prmvia58252.2023.00046","url":null,"abstract":"SLAM (simultaneous localization and mapping), will further known as synchronous localization and mapping, is a technology that is used to tackle the issue of localization and map building while a robot travels in an unfamiliar environment. Traditional SLAM relies on point features to estimate camera pose, which makes it difficult to extract enough point features in low-texture scenes. When the camera shakes violently or rotates too fast, the robustness of a point-based SLAM system is poor. Aiming at the problem of poor robustness of the existing visual SLAM (synchronous localization and mapping technology) system, based on the ORB-SLAM3 framework, the point feature extractor is replaced with a self-supervised deep neural network, and a matching filtering algorithm based on threshold and motion statistics is proposed to eliminate point mismatch, this significantly accelerates the system’s real- time and accuracy. Likewise, linear activities are integrated into the front-end information extraction, a linear feature extraction model is established, approximation linear features are merged and processed, and the linear feature description and mismatching eradication process are simplified. Finally, the weight allocation idea is introduced into the construction of the point and line error model, and the weight of the point and line is reasonably allocated according to the richness of the scene. Experiments on absolute error trajectory on the TUM dataset emphasize that the revised algorithm increased efficiency and stability when compared to the ORB-SLAM3 system.","PeriodicalId":221346,"journal":{"name":"2023 International Conference on Pattern Recognition, Machine Vision and Intelligent Algorithms (PRMVIA)","volume":"5 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116971010","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 noise of the received ECG signal, a extraction method of ECG signal based on singular value selection and wavelet is proposed. The singular value decomposition on the ECG signal is performed firstly, and the ECG signal component corresponding to each singular value is obtained. Then the signal component corresponding to the maximum singular value is used to calculate cross-correlation coefficients with other components. The cumulative contribution rate of singular values is combined to determine the number of singular values for ECG signal reconstruction. The wavelet threshold de-noising method is used to de-noise the final determined signal components. Finally, the de-noising ECG signal is obtained by reconstructing the signal components. The experimental results show that the method can suppress noise and extract signal effectively, and it has a good noise reduction effect compared with wavelet threshold method.
{"title":"ECG Signal Extraction Method Based on Singular Value Selection and Wavelet","authors":"Fuyu Luo, Xue Han, Zihao Zhang, Ruigang Li, Huixi Wang, Fanrong Kong","doi":"10.1109/PRMVIA58252.2023.00052","DOIUrl":"https://doi.org/10.1109/PRMVIA58252.2023.00052","url":null,"abstract":"Aiming at the noise of the received ECG signal, a extraction method of ECG signal based on singular value selection and wavelet is proposed. The singular value decomposition on the ECG signal is performed firstly, and the ECG signal component corresponding to each singular value is obtained. Then the signal component corresponding to the maximum singular value is used to calculate cross-correlation coefficients with other components. The cumulative contribution rate of singular values is combined to determine the number of singular values for ECG signal reconstruction. The wavelet threshold de-noising method is used to de-noise the final determined signal components. Finally, the de-noising ECG signal is obtained by reconstructing the signal components. The experimental results show that the method can suppress noise and extract signal effectively, and it has a good noise reduction effect compared with wavelet threshold method.","PeriodicalId":221346,"journal":{"name":"2023 International Conference on Pattern Recognition, Machine Vision and Intelligent Algorithms (PRMVIA)","volume":"65 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117249494","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 : 2023-03-01DOI: 10.1109/PRMVIA58252.2023.00053
Hongjie Zhou
Few-shot semantic segmentation (FSS) requires only few labeled samples to achieve good segmentation performance and thus has received extensive attention. However, existing FFS methods usually adopt a simple convolutional structure as the backbone, which suffers from poor feature extraction ability. In order to address this issue, a novel few-shot segmentation network based on dual-branch feature extraction (DFESN) is proposed. First, an attention-enhanced ResNet is used as the local feature extraction branch. Specifically, we in-corporate channel attention operations into each building block of ResNet to model the importance among channels, which enables DFESN to learn important class information for the segmentation task. Besides, we introduce a Vision Transformer as the global feature extraction branch. This branch leverages the multi-head self-attention mechanism in Vision Transformer to model the global dependencies of support and query image features, further enhancing the feature extraction capabilities of DFESN. We conduct experiments on the PASCAL-5i dataset and demonstrate the superiority of our DFESN.
{"title":"Few-Shot Semantic Segmentation Based on Dual-Branch Feature Extraction","authors":"Hongjie Zhou","doi":"10.1109/PRMVIA58252.2023.00053","DOIUrl":"https://doi.org/10.1109/PRMVIA58252.2023.00053","url":null,"abstract":"Few-shot semantic segmentation (FSS) requires only few labeled samples to achieve good segmentation performance and thus has received extensive attention. However, existing FFS methods usually adopt a simple convolutional structure as the backbone, which suffers from poor feature extraction ability. In order to address this issue, a novel few-shot segmentation network based on dual-branch feature extraction (DFESN) is proposed. First, an attention-enhanced ResNet is used as the local feature extraction branch. Specifically, we in-corporate channel attention operations into each building block of ResNet to model the importance among channels, which enables DFESN to learn important class information for the segmentation task. Besides, we introduce a Vision Transformer as the global feature extraction branch. This branch leverages the multi-head self-attention mechanism in Vision Transformer to model the global dependencies of support and query image features, further enhancing the feature extraction capabilities of DFESN. We conduct experiments on the PASCAL-5i dataset and demonstrate the superiority of our DFESN.","PeriodicalId":221346,"journal":{"name":"2023 International Conference on Pattern Recognition, Machine Vision and Intelligent Algorithms (PRMVIA)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124267186","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}
Tuberculosis and diabetes mellitus are highly prevalent clinical conditions worldwide, and the mortality rate of tuberculosis is high; when diabetes mellitus is combined with tuberculosis, the interaction between the two can lead to a vicious cycle, posing a serious threat to the physical and mental health and life safety of patients, especially in developing regions where medical resources are scarce. In this paper, We trained several deep learning algorithm models based on YOLOv5, YOLOv8x, Faster R- CNN and Mask R-CNN with 4 types of lesion features commonly found in 1024 images, from which we selected the algorithm with the best automatic feature recognition effect and optimized the model to further improve the recognition efficiency. Combining the complexity of lesion features and experimental results, we propose a YOLOv8x model based on RepEca network structure and ESE attention mechanism, which is more effective than the original YOLOv8x in application, with an F1 metric value of 71.19%, and can better identify lesion features in images, assisting clinicians to improve the diagnostic accuracy and treatment effect.
{"title":"Identification Of Imaging Features Of Diabetes Mellitus And Tuberculosis Based On YOLOv8x Model Combined With RepEca Network Structure","authors":"Wenjun Li, Linjun Jiang, Zezhou Zhu, Yanfan Li, Hua Peng, Diqing Liang, Hongzhong Yang, Weijun Liang","doi":"10.1109/prmvia58252.2023.00032","DOIUrl":"https://doi.org/10.1109/prmvia58252.2023.00032","url":null,"abstract":"Tuberculosis and diabetes mellitus are highly prevalent clinical conditions worldwide, and the mortality rate of tuberculosis is high; when diabetes mellitus is combined with tuberculosis, the interaction between the two can lead to a vicious cycle, posing a serious threat to the physical and mental health and life safety of patients, especially in developing regions where medical resources are scarce. In this paper, We trained several deep learning algorithm models based on YOLOv5, YOLOv8x, Faster R- CNN and Mask R-CNN with 4 types of lesion features commonly found in 1024 images, from which we selected the algorithm with the best automatic feature recognition effect and optimized the model to further improve the recognition efficiency. Combining the complexity of lesion features and experimental results, we propose a YOLOv8x model based on RepEca network structure and ESE attention mechanism, which is more effective than the original YOLOv8x in application, with an F1 metric value of 71.19%, and can better identify lesion features in images, assisting clinicians to improve the diagnostic accuracy and treatment effect.","PeriodicalId":221346,"journal":{"name":"2023 International Conference on Pattern Recognition, Machine Vision and Intelligent Algorithms (PRMVIA)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130444002","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 : 2023-03-01DOI: 10.1109/PRMVIA58252.2023.00049
Zhimin Zhang, Zhaolian Wang, Chenglong Zhang, Xiaoli Yang, Xiaopeng Ma
Photoacoustic tomography often faces problems such as incomplete data and noise, which affect the quality of reconstructed images. Model-based photoacoustic image reconstruction is an ill-posed inverse problem, which usually needs to introduce the regularization term as the prior constraint. In this paper, we propose a novel model-based regularization framework for photoacoustic image reconstruction, which utilizes the group sparsity property of photoacoustic images as prior information and combines total variation regularization to effectively suppress image artifacts and recover the missing signal data during sparse sampling. Numerical simulation results show that the proposed algorithm not only improves the accuracy of photoacoustic reconstruction under sparse sampling but also improves the calculation speed.
{"title":"Sparse sampling photoacoustic reconstruction with group sparse dictionary learning","authors":"Zhimin Zhang, Zhaolian Wang, Chenglong Zhang, Xiaoli Yang, Xiaopeng Ma","doi":"10.1109/PRMVIA58252.2023.00049","DOIUrl":"https://doi.org/10.1109/PRMVIA58252.2023.00049","url":null,"abstract":"Photoacoustic tomography often faces problems such as incomplete data and noise, which affect the quality of reconstructed images. Model-based photoacoustic image reconstruction is an ill-posed inverse problem, which usually needs to introduce the regularization term as the prior constraint. In this paper, we propose a novel model-based regularization framework for photoacoustic image reconstruction, which utilizes the group sparsity property of photoacoustic images as prior information and combines total variation regularization to effectively suppress image artifacts and recover the missing signal data during sparse sampling. Numerical simulation results show that the proposed algorithm not only improves the accuracy of photoacoustic reconstruction under sparse sampling but also improves the calculation speed.","PeriodicalId":221346,"journal":{"name":"2023 International Conference on Pattern Recognition, Machine Vision and Intelligent Algorithms (PRMVIA)","volume":"4 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130700691","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 : 2023-03-01DOI: 10.1109/prmvia58252.2023.00014
Yongchang Zhang, Yue Guo, Hanbing Niu, Bo Zhang, Yun Cao, Wenhao He
Achieving robust 3D object detection by fusing images and point clouds remains challenging. In this paper, we propose a novel 3D object detector (SimpleFusion) that enables simple and efficient multi-sensor fusion. Our main motivation is to boost feature extraction from a single modality and fuse them into a unified space. Specifically, we build a new visual 3D object detector in the camera stream that leverages point cloud supervision for more accurate depth prediction; in the lidar stream, we introduce a robust 3D object detector that utilizes multi-view and multi-scale features to overcome the sparsity of point clouds. Finally, we propose a dynamic fusion module to focus on more confident features and achieve accurate 3D object detection based on dynamic weights. Our method has been evaluated on the nuScenes dataset, and the experimental results indicate that it outperforms other state-of-the-art methods by a significant margin.
{"title":"SimpleFusion: 3D Object Detection by Fusing RGB Images and Point Clouds","authors":"Yongchang Zhang, Yue Guo, Hanbing Niu, Bo Zhang, Yun Cao, Wenhao He","doi":"10.1109/prmvia58252.2023.00014","DOIUrl":"https://doi.org/10.1109/prmvia58252.2023.00014","url":null,"abstract":"Achieving robust 3D object detection by fusing images and point clouds remains challenging. In this paper, we propose a novel 3D object detector (SimpleFusion) that enables simple and efficient multi-sensor fusion. Our main motivation is to boost feature extraction from a single modality and fuse them into a unified space. Specifically, we build a new visual 3D object detector in the camera stream that leverages point cloud supervision for more accurate depth prediction; in the lidar stream, we introduce a robust 3D object detector that utilizes multi-view and multi-scale features to overcome the sparsity of point clouds. Finally, we propose a dynamic fusion module to focus on more confident features and achieve accurate 3D object detection based on dynamic weights. Our method has been evaluated on the nuScenes dataset, and the experimental results indicate that it outperforms other state-of-the-art methods by a significant margin.","PeriodicalId":221346,"journal":{"name":"2023 International Conference on Pattern Recognition, Machine Vision and Intelligent Algorithms (PRMVIA)","volume":"36 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128379902","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：