Pub Date : 2022-12-09DOI: 10.1109/ACAIT56212.2022.10137805
Huan Wu, Feng Jiang, Zehua Zhao, Liang Tao, Lin Peng, Wenxun Han, M. Gao
Water quality prediction is widely used in many aspects of water pollution. In recent years, researchers have begun using deep learning models for prediction, significantly improving prediction performance. However, these methods usually aim single-prediction task. They are not good at linking multiple prediction indicators of water quality, and are difficult to train at water quality monitoring stations with small samples. For this reason, we explore a model based on multi-indicators relationship learning and knowledge transfer for the prediction in a small sample scenario. Firstly, we propose a water quality multi-indicator gated implicit variable parameter sharing model MGH to extract the common characteristics, individual characteristics, and correlation of water quality multi-indicators at multi-sample sites. Then, we combine two migration methods to carry out the water quality prediction model in the small sample area to migrate the trained model parameters or sample distribution knowledge to the small sample area. The experimental results show that our multi-indicator relationship learning and knowledge transfer model can achieve better prediction accuracy. We also verified the role of the model in ensuring the prediction effect of small sample monitor stations. The proposed model provides data support for water quality managers to understand the changes in water quality indicators in advance and make corresponding water quality management decisions.
{"title":"Water Quality Prediction Based on Multi-Indicator Relation Learning and Knowledge Transfer","authors":"Huan Wu, Feng Jiang, Zehua Zhao, Liang Tao, Lin Peng, Wenxun Han, M. Gao","doi":"10.1109/ACAIT56212.2022.10137805","DOIUrl":"https://doi.org/10.1109/ACAIT56212.2022.10137805","url":null,"abstract":"Water quality prediction is widely used in many aspects of water pollution. In recent years, researchers have begun using deep learning models for prediction, significantly improving prediction performance. However, these methods usually aim single-prediction task. They are not good at linking multiple prediction indicators of water quality, and are difficult to train at water quality monitoring stations with small samples. For this reason, we explore a model based on multi-indicators relationship learning and knowledge transfer for the prediction in a small sample scenario. Firstly, we propose a water quality multi-indicator gated implicit variable parameter sharing model MGH to extract the common characteristics, individual characteristics, and correlation of water quality multi-indicators at multi-sample sites. Then, we combine two migration methods to carry out the water quality prediction model in the small sample area to migrate the trained model parameters or sample distribution knowledge to the small sample area. The experimental results show that our multi-indicator relationship learning and knowledge transfer model can achieve better prediction accuracy. We also verified the role of the model in ensuring the prediction effect of small sample monitor stations. The proposed model provides data support for water quality managers to understand the changes in water quality indicators in advance and make corresponding water quality management decisions.","PeriodicalId":398228,"journal":{"name":"2022 6th Asian Conference on Artificial Intelligence Technology (ACAIT)","volume":"60 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125659267","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 : 2022-12-09DOI: 10.1109/ACAIT56212.2022.10137903
Zhenyu Xin
To further reduce the engineering cost of construction project, an engineering cost management prediction model of construction project based on BIM technology and BP neural network is proposed. Among them, the construction project index is taken as the input, and the BIM technology is used to calculate the project quantity. Then the bill of quantity is taken as the input of BP neural network, so as to predict the cost of the engineering cost. The results show that after the BP neural network is trained in MATLB software. Moreover, the fitting effect of the prediction model is significantly improved. The actual prediction shows that the predicted value of meter cost using this model is very close to the actual value of meter cost, and the maximum error between them is only 266. It shows that using the proposed model can improve the accuracy of engineering cost prediction of construction project, so as to further reduce the cost of the construction project.
{"title":"Research on Engineering Cost Management of Construction Project Based on BIM Technology and BP Neural Network","authors":"Zhenyu Xin","doi":"10.1109/ACAIT56212.2022.10137903","DOIUrl":"https://doi.org/10.1109/ACAIT56212.2022.10137903","url":null,"abstract":"To further reduce the engineering cost of construction project, an engineering cost management prediction model of construction project based on BIM technology and BP neural network is proposed. Among them, the construction project index is taken as the input, and the BIM technology is used to calculate the project quantity. Then the bill of quantity is taken as the input of BP neural network, so as to predict the cost of the engineering cost. The results show that after the BP neural network is trained in MATLB software. Moreover, the fitting effect of the prediction model is significantly improved. The actual prediction shows that the predicted value of meter cost using this model is very close to the actual value of meter cost, and the maximum error between them is only 266. It shows that using the proposed model can improve the accuracy of engineering cost prediction of construction project, so as to further reduce the cost of the construction project.","PeriodicalId":398228,"journal":{"name":"2022 6th Asian Conference on Artificial Intelligence Technology (ACAIT)","volume":"18 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130161600","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 : 2022-12-09DOI: 10.1109/ACAIT56212.2022.10137781
Xiangbo Diao, Yong Xu
With the massive installation of cameras in cities, the requirement of equipment in an urban safety system has been basically satisfied. As a result, an available intelligent video-based safety system is very important. Video-based violence recognition method, which plays an absolutely important role in urban safety, seems to be a useful component of this system. However, a large part of existing violence recognition methods encounter the problems of low efficiency and inaccuracy owing to enormous challenges in accurately identifying the various violence events. In this paper, we propose a violence recognition algorithm based on the SlowFast model. In order to obtain a good performance, we modify the SlowFast model by both improving the convolutional structure and inserting plug-and-play modules. These two schemes can improve the speed and accuracy of violence recognition. The improved model achieves a high accuracy on several publicly available violence datasets surpassing some previous works and can be applied to violence detection in real-world scenarios, which is beneficial to fight crime and maintain social security.
{"title":"A SlowFast-Based Violence Recognition Method","authors":"Xiangbo Diao, Yong Xu","doi":"10.1109/ACAIT56212.2022.10137781","DOIUrl":"https://doi.org/10.1109/ACAIT56212.2022.10137781","url":null,"abstract":"With the massive installation of cameras in cities, the requirement of equipment in an urban safety system has been basically satisfied. As a result, an available intelligent video-based safety system is very important. Video-based violence recognition method, which plays an absolutely important role in urban safety, seems to be a useful component of this system. However, a large part of existing violence recognition methods encounter the problems of low efficiency and inaccuracy owing to enormous challenges in accurately identifying the various violence events. In this paper, we propose a violence recognition algorithm based on the SlowFast model. In order to obtain a good performance, we modify the SlowFast model by both improving the convolutional structure and inserting plug-and-play modules. These two schemes can improve the speed and accuracy of violence recognition. The improved model achieves a high accuracy on several publicly available violence datasets surpassing some previous works and can be applied to violence detection in real-world scenarios, which is beneficial to fight crime and maintain social security.","PeriodicalId":398228,"journal":{"name":"2022 6th Asian Conference on Artificial Intelligence Technology (ACAIT)","volume":"74 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128820341","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}
Recommendation system is an significant study goal in the realm of information filtering system. The recommender system predicts the items that users are interested in depended on the user’s past operational data. Dynamic behavior sequence can be extracted by self-attention mechanism, and most models assume that the interaction history is regarded as an ordered sequence without considering the time interval between each interaction. Our input entities and items are both interconnected and highly correlated in the knowledge graph and recommendation modules respectively. This paper fuses them to recommend to users, and proposes a multi-task feature learning recommendation model that fuses time interval and knowledge graph, explicitly modeling interaction timestamps within a sequential modeling framework, and fused with knowledge graph embedding (KGE) to assist with the recommended task. The experimental results show that on the real dataset MovieLens1M, the AUC, ACC, Precision and Recall indicators are used for evaluation, and the proposed model is better than the mainstream benchmark models.
{"title":"Research on Recommendation Method Based on Fusion of Knowledge Graph and Behavioral Time Interval","authors":"Hao Lin, Caimao Li, Shaofan Chen, Qiu-Shuang Chen, Yuquan Hou, Hao Li","doi":"10.1109/ACAIT56212.2022.10137803","DOIUrl":"https://doi.org/10.1109/ACAIT56212.2022.10137803","url":null,"abstract":"Recommendation system is an significant study goal in the realm of information filtering system. The recommender system predicts the items that users are interested in depended on the user’s past operational data. Dynamic behavior sequence can be extracted by self-attention mechanism, and most models assume that the interaction history is regarded as an ordered sequence without considering the time interval between each interaction. Our input entities and items are both interconnected and highly correlated in the knowledge graph and recommendation modules respectively. This paper fuses them to recommend to users, and proposes a multi-task feature learning recommendation model that fuses time interval and knowledge graph, explicitly modeling interaction timestamps within a sequential modeling framework, and fused with knowledge graph embedding (KGE) to assist with the recommended task. The experimental results show that on the real dataset MovieLens1M, the AUC, ACC, Precision and Recall indicators are used for evaluation, and the proposed model is better than the mainstream benchmark models.","PeriodicalId":398228,"journal":{"name":"2022 6th Asian Conference on Artificial Intelligence Technology (ACAIT)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127708044","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 : 2022-12-09DOI: 10.1109/ACAIT56212.2022.10137925
haocheng gao, Xin Xu, Changxin Zhang, Xing Zhou
In recent years, learning-based approaches for solving combinational optimization problems have received increasing research interest. However, it is still challenging to solve multi-objective optimization problems (MOPs). In this paper, we proposed a bidirectional parameter transfer attention-based reinforcement learning approach for solving bi-objective traveling salesman problem (BOTSP), which is based on dynamic context attention neural network trained by the rollout reinforce algorithm. Specifically, BOTSP is decomposed into a series of static sub-tasks at first, then, bidirectional parameter transfer methods are proposed for training each subproblem sequentially. Once the model has been learned, Pareto optimal solutions can be obtained on different scale problem instances. Extensive experiments on BOTSP were conducted to illustrate the effectiveness and advantages of the proposed approach. Compared with several algorithms, our proposed method achieves the state-of-the-art performance in hypervolume and inference efficiency. In particular, our method is suitable for different scale problem instances without extra learning, and experimental results demonstrate it realizes powerful generalization ability across tasks.
{"title":"A Bidirectional Parameter Transfer Reinforcement Learning Approach for Bi-Objectives Traveling Salesman Problem","authors":"haocheng gao, Xin Xu, Changxin Zhang, Xing Zhou","doi":"10.1109/ACAIT56212.2022.10137925","DOIUrl":"https://doi.org/10.1109/ACAIT56212.2022.10137925","url":null,"abstract":"In recent years, learning-based approaches for solving combinational optimization problems have received increasing research interest. However, it is still challenging to solve multi-objective optimization problems (MOPs). In this paper, we proposed a bidirectional parameter transfer attention-based reinforcement learning approach for solving bi-objective traveling salesman problem (BOTSP), which is based on dynamic context attention neural network trained by the rollout reinforce algorithm. Specifically, BOTSP is decomposed into a series of static sub-tasks at first, then, bidirectional parameter transfer methods are proposed for training each subproblem sequentially. Once the model has been learned, Pareto optimal solutions can be obtained on different scale problem instances. Extensive experiments on BOTSP were conducted to illustrate the effectiveness and advantages of the proposed approach. Compared with several algorithms, our proposed method achieves the state-of-the-art performance in hypervolume and inference efficiency. In particular, our method is suitable for different scale problem instances without extra learning, and experimental results demonstrate it realizes powerful generalization ability across tasks.","PeriodicalId":398228,"journal":{"name":"2022 6th Asian Conference on Artificial Intelligence Technology (ACAIT)","volume":"33 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127810262","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 : 2022-12-09DOI: 10.1109/ACAIT56212.2022.10137820
S. Deng, Zhiyuan Du, Wenhao Shen, Zhinan Gao, Yifan Wang, Lingyun Zhu
RoboCup is a multi-robot collaborative soccer competition platform. A recognized problem in applying artificial intelligence technology is accomplishing complex tasks of multi-robot cooperation in the match according to changing environmental conditions. The core part of the decision-making system determines whether a robot team can cooperate reasonably. This paper optimizes the finite state machine decision-making system based on the B-human framework by perfecting the decision tree system, subdivision site zoning, improving the communication within the team, and designing keyframe actions. Team formations, robot roles, and execution actions change dynamically depending on different competitive states in this system. This decision system works well in SimRobot simulations, showing the strategic advantages of real-time, distributed features of multi-robot collaboration in real competitions.
{"title":"Multi-Robot Cooperative Defense Strategy in RoboCup Standard Platform League","authors":"S. Deng, Zhiyuan Du, Wenhao Shen, Zhinan Gao, Yifan Wang, Lingyun Zhu","doi":"10.1109/ACAIT56212.2022.10137820","DOIUrl":"https://doi.org/10.1109/ACAIT56212.2022.10137820","url":null,"abstract":"RoboCup is a multi-robot collaborative soccer competition platform. A recognized problem in applying artificial intelligence technology is accomplishing complex tasks of multi-robot cooperation in the match according to changing environmental conditions. The core part of the decision-making system determines whether a robot team can cooperate reasonably. This paper optimizes the finite state machine decision-making system based on the B-human framework by perfecting the decision tree system, subdivision site zoning, improving the communication within the team, and designing keyframe actions. Team formations, robot roles, and execution actions change dynamically depending on different competitive states in this system. This decision system works well in SimRobot simulations, showing the strategic advantages of real-time, distributed features of multi-robot collaboration in real competitions.","PeriodicalId":398228,"journal":{"name":"2022 6th Asian Conference on Artificial Intelligence Technology (ACAIT)","volume":"3 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127825756","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 : 2022-12-09DOI: 10.1109/ACAIT56212.2022.10137783
Ning Zhang, Yisheng Miao, Huarui Wu, Xiang Sun, Huaji Zhu
Judging the seedling age of cabbage seedlings at the seedling stage is helpful for the production management of cabbage seedlings, and is of great significance for guiding the fertilization amount of seedling production operations. In order to realize the accurate judgment of the number of cabbage leaves in the complex environment of the nursery greenhouse, in view of the problem that the target size of the leaves of the cabbage seedlings is small and difficult to identify, a method for counting the leaves of the cabbage seedlings based on instance segmentation was proposed. The model structure is based on the Mask R-CNN instance segmentation model, using Resnet50 as the feature extractor, adding deformable convolution to improve the feature extraction capability of the model, and selecting the category cross entropy as the loss function. The model is verified on the cabbage seedling data set constructed by self-collection. The proposed method is better than yoloV3 and FPN. The coefficient of determination, root mean square error and mean absolute error of the model trained in this paper reach 0.93, 6.24, and 4.63, compared with yoloV3 and FPN. The original network, the counting accuracy is improved. The method can accurately identify the number of leaves in each growth stage of cabbage seedlings, and provide an effective theoretical basis for the informatization of facility cabbage seedling production.
{"title":"A Method for Counting Leaves of Cabbage Seedlings Based on Instance Segmentation","authors":"Ning Zhang, Yisheng Miao, Huarui Wu, Xiang Sun, Huaji Zhu","doi":"10.1109/ACAIT56212.2022.10137783","DOIUrl":"https://doi.org/10.1109/ACAIT56212.2022.10137783","url":null,"abstract":"Judging the seedling age of cabbage seedlings at the seedling stage is helpful for the production management of cabbage seedlings, and is of great significance for guiding the fertilization amount of seedling production operations. In order to realize the accurate judgment of the number of cabbage leaves in the complex environment of the nursery greenhouse, in view of the problem that the target size of the leaves of the cabbage seedlings is small and difficult to identify, a method for counting the leaves of the cabbage seedlings based on instance segmentation was proposed. The model structure is based on the Mask R-CNN instance segmentation model, using Resnet50 as the feature extractor, adding deformable convolution to improve the feature extraction capability of the model, and selecting the category cross entropy as the loss function. The model is verified on the cabbage seedling data set constructed by self-collection. The proposed method is better than yoloV3 and FPN. The coefficient of determination, root mean square error and mean absolute error of the model trained in this paper reach 0.93, 6.24, and 4.63, compared with yoloV3 and FPN. The original network, the counting accuracy is improved. The method can accurately identify the number of leaves in each growth stage of cabbage seedlings, and provide an effective theoretical basis for the informatization of facility cabbage seedling production.","PeriodicalId":398228,"journal":{"name":"2022 6th Asian Conference on Artificial Intelligence Technology (ACAIT)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130445162","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 : 2022-12-09DOI: 10.1109/ACAIT56212.2022.10137954
Kun-Yi Chen, Suqin Guo, Han Li, Peishu Wu, Nianyin Zeng
Remote sensing technique has played important roles in various fields like urban planning and military reconnaissance, however, due to remote sensing images (RSI) have the unique characteristics of complicated background, densely distribution of targets with varying scales, etc., it remains a challenging work to apply popular object detection algorithms for RSI analysis. In this paper, an improved Yolo-v3 (Im-Yolo) model is developed with enhanced feature learning ability, which can better adapt to handling RSI. In particular, residual convolution and path aggregation are employed so as to effectively enhance the multi-scale feature extraction and semantic-detail information fusion ability of Im-Yolo. Experiments on two challenging remote sensing detection databases have sufficiently demonstrated the reliability and superiority of proposed Im-Yolo on both detection accuracy and inference speed in comparison to the baseline model Yolo-v3. Im-Yolo is proven a competent method for handling RSI with satisfactory performances even in complicated scenarios, which can provide experiences to design RSI-oriented object detection algorithms.
{"title":"Improved Yolo-v3 Model with Enhanced Feature Learning for Remote Sensing Image Analysis","authors":"Kun-Yi Chen, Suqin Guo, Han Li, Peishu Wu, Nianyin Zeng","doi":"10.1109/ACAIT56212.2022.10137954","DOIUrl":"https://doi.org/10.1109/ACAIT56212.2022.10137954","url":null,"abstract":"Remote sensing technique has played important roles in various fields like urban planning and military reconnaissance, however, due to remote sensing images (RSI) have the unique characteristics of complicated background, densely distribution of targets with varying scales, etc., it remains a challenging work to apply popular object detection algorithms for RSI analysis. In this paper, an improved Yolo-v3 (Im-Yolo) model is developed with enhanced feature learning ability, which can better adapt to handling RSI. In particular, residual convolution and path aggregation are employed so as to effectively enhance the multi-scale feature extraction and semantic-detail information fusion ability of Im-Yolo. Experiments on two challenging remote sensing detection databases have sufficiently demonstrated the reliability and superiority of proposed Im-Yolo on both detection accuracy and inference speed in comparison to the baseline model Yolo-v3. Im-Yolo is proven a competent method for handling RSI with satisfactory performances even in complicated scenarios, which can provide experiences to design RSI-oriented object detection algorithms.","PeriodicalId":398228,"journal":{"name":"2022 6th Asian Conference on Artificial Intelligence Technology (ACAIT)","volume":"3 4 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130483023","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 : 2022-12-09DOI: 10.1109/ACAIT56212.2022.10137985
Wei Li, Qingzheng Xu, Xinyu Gao, Lei Wang
Multi-factorial optimization (MFO) is a popular optimization mechanism recently, which aims to optimize multiple tasks in a single run. Generally, the multi-factorial optimization algorithm uses the random learning strategy to generate offspring, which leads to an inefficient exploration ability of the algorithm. To address this problem, this paper proposed a simple and effective strategy called team learning strategy (TLS). The proposed learning strategy divided the population into excellent team and ordinary team. Different teams employ different learning strategies. Moreover, nine MFO problems were used to assess the effectiveness of the proposed strategy. The experimental results show that the team learning strategy can improve the performance of the algorithm.
{"title":"Evolutionary Multitasking Based on Team Learning Strategy","authors":"Wei Li, Qingzheng Xu, Xinyu Gao, Lei Wang","doi":"10.1109/ACAIT56212.2022.10137985","DOIUrl":"https://doi.org/10.1109/ACAIT56212.2022.10137985","url":null,"abstract":"Multi-factorial optimization (MFO) is a popular optimization mechanism recently, which aims to optimize multiple tasks in a single run. Generally, the multi-factorial optimization algorithm uses the random learning strategy to generate offspring, which leads to an inefficient exploration ability of the algorithm. To address this problem, this paper proposed a simple and effective strategy called team learning strategy (TLS). The proposed learning strategy divided the population into excellent team and ordinary team. Different teams employ different learning strategies. Moreover, nine MFO problems were used to assess the effectiveness of the proposed strategy. The experimental results show that the team learning strategy can improve the performance of the algorithm.","PeriodicalId":398228,"journal":{"name":"2022 6th Asian Conference on Artificial Intelligence Technology (ACAIT)","volume":"98 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124545954","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 : 2022-12-09DOI: 10.1109/ACAIT56212.2022.10137961
Zezhong Xu, Cheng Qian, Qingxiang You, Feng Wu
A fast method for lanes detection is proposed to deal with worn lane markings. Lanes are detected based on regional pixels rather than edge points in this paper by concerning that the worn or faded lane markings lead to disruption to edge extraction. After pixels have voted, the local maximum is searched and a peak region is defined in the Hough space. The voting of a column in the peak region is considered as a stochastic variable, and the statistical characteristics are computed. The statistical variances are used to fit a quadratic function. The direction parameter of the lane marking is determined by the minimization of the fitted quadratic function. The statistical means are used to fit a linear function. The position parameter of the lane marking is computed using the interpolation technique. Lane tracking is implemented with lower computation cost by defining the peak region based on the lane parameters detected in previous frame. The experimental results show that the proposed method can detect effectively the lane markings even in presence of seriously worn roads. The computation time is less than 2ms for a road image.
{"title":"Fast Detection and Tracking of Worn Lane Markings","authors":"Zezhong Xu, Cheng Qian, Qingxiang You, Feng Wu","doi":"10.1109/ACAIT56212.2022.10137961","DOIUrl":"https://doi.org/10.1109/ACAIT56212.2022.10137961","url":null,"abstract":"A fast method for lanes detection is proposed to deal with worn lane markings. Lanes are detected based on regional pixels rather than edge points in this paper by concerning that the worn or faded lane markings lead to disruption to edge extraction. After pixels have voted, the local maximum is searched and a peak region is defined in the Hough space. The voting of a column in the peak region is considered as a stochastic variable, and the statistical characteristics are computed. The statistical variances are used to fit a quadratic function. The direction parameter of the lane marking is determined by the minimization of the fitted quadratic function. The statistical means are used to fit a linear function. The position parameter of the lane marking is computed using the interpolation technique. Lane tracking is implemented with lower computation cost by defining the peak region based on the lane parameters detected in previous frame. The experimental results show that the proposed method can detect effectively the lane markings even in presence of seriously worn roads. The computation time is less than 2ms for a road image.","PeriodicalId":398228,"journal":{"name":"2022 6th Asian Conference on Artificial Intelligence Technology (ACAIT)","volume":"233 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122533850","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：