Pub Date : 2025-01-14DOI: 10.1109/TITS.2024.3518293
{"title":"IEEE Intelligent Transportation Systems Society Information","authors":"","doi":"10.1109/TITS.2024.3518293","DOIUrl":"https://doi.org/10.1109/TITS.2024.3518293","url":null,"abstract":"","PeriodicalId":13416,"journal":{"name":"IEEE Transactions on Intelligent Transportation Systems","volume":"26 1","pages":"C3-C3"},"PeriodicalIF":7.9,"publicationDate":"2025-01-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10841913","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142976043","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-14DOI: 10.1109/TITS.2024.3518135
Simona Sacone
“Scanning the Issue.“
“扫描问题”。”
{"title":"Scanning the Issue","authors":"Simona Sacone","doi":"10.1109/TITS.2024.3518135","DOIUrl":"https://doi.org/10.1109/TITS.2024.3518135","url":null,"abstract":"“Scanning the Issue.“","PeriodicalId":13416,"journal":{"name":"IEEE Transactions on Intelligent Transportation Systems","volume":"26 1","pages":"3-21"},"PeriodicalIF":7.9,"publicationDate":"2025-01-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10841924","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142992917","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-14DOI: 10.1109/TITS.2024.3503496
Maurice Kolff;Joost Venrooij;Elena Arcidiacono;Daan M. Pool;Max Mulder
This paper presents a three-step validation approach for subjective rating predictions of driving simulator motion incongruences based on objective mismatches between reference vehicle and simulator motion. This approach relies on using high-resolution rating predictions of open-loop driving (participants being driven) for ratings of motion in closed-loop driving (participants driving themselves). A driving simulator experiment in an urban scenario is described, of which the rating data of 36 participants was recorded and analyzed. In the experiment’s first phase, participants actively drove themselves (i.e., closed-loop). By recording the drives of the participants and playing these back to themselves (open-loop) in the second phase, participants experienced the same motion in both phases. Participants rated the motion after each maneuver and at the end of each drive. In the third phase they again drove open-loop, but rated the motion continuously, only possible in open-loop driving. Results show that a rating model, acquired through a different experiment, can well predict the measured continuous ratings. Second, the maximum of the measured continuous ratings correlates to both the maneuver-based ($rho =0.94$ ) and overall ($rho =0.69$ ) ratings, allowing for predictions of both rating types based on the continuous rating model. Third, using Bayesian statistics it is then shown that both the maneuver-based and overall ratings between the closed-loop and open-loop drives are equivalent. This allows for predictions of maneuver-based and overall ratings using the high-resolution continuous rating models. These predictions can be used as an accurate trade-off method of motion cueing settings of future closed-loop driving simulator experiments.
{"title":"Predicting Motion Incongruence Ratings in Closed- and Open-Loop Urban Driving Simulation","authors":"Maurice Kolff;Joost Venrooij;Elena Arcidiacono;Daan M. Pool;Max Mulder","doi":"10.1109/TITS.2024.3503496","DOIUrl":"https://doi.org/10.1109/TITS.2024.3503496","url":null,"abstract":"This paper presents a three-step validation approach for subjective rating predictions of driving simulator motion incongruences based on objective mismatches between reference vehicle and simulator motion. This approach relies on using high-resolution rating predictions of open-loop driving (participants being driven) for ratings of motion in closed-loop driving (participants driving themselves). A driving simulator experiment in an urban scenario is described, of which the rating data of 36 participants was recorded and analyzed. In the experiment’s first phase, participants actively drove themselves (i.e., closed-loop). By recording the drives of the participants and playing these back to themselves (open-loop) in the second phase, participants experienced the same motion in both phases. Participants rated the motion after each maneuver and at the end of each drive. In the third phase they again drove open-loop, but rated the motion continuously, only possible in open-loop driving. Results show that a rating model, acquired through a different experiment, can well predict the measured continuous ratings. Second, the maximum of the measured continuous ratings correlates to both the maneuver-based (<inline-formula> <tex-math>$rho =0.94$ </tex-math></inline-formula>) and overall (<inline-formula> <tex-math>$rho =0.69$ </tex-math></inline-formula>) ratings, allowing for predictions of both rating types based on the continuous rating model. Third, using Bayesian statistics it is then shown that both the maneuver-based and overall ratings between the closed-loop and open-loop drives are equivalent. This allows for predictions of maneuver-based and overall ratings using the high-resolution continuous rating models. These predictions can be used as an accurate trade-off method of motion cueing settings of future closed-loop driving simulator experiments.","PeriodicalId":13416,"journal":{"name":"IEEE Transactions on Intelligent Transportation Systems","volume":"26 1","pages":"517-528"},"PeriodicalIF":7.9,"publicationDate":"2025-01-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142976138","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-12-16DOI: 10.1109/TITS.2024.3505237
Bingyan Cui;Hao Wang
Uncertainty in climate change poses challenges in obtaining accurate and reliable prediction models for future pavement performance. This study aimed to develop an advanced prediction model specifically for flexible pavements, incorporating uncertainty quantification through a Bayesian Neural Network (BNN). Focusing on predicting the International Roughness Index (IRI) and rut depth of asphalt pavement, BNN model was applied to different climate regions, using long-term pavement performance (LTPP) data from 1989 to 2021. The Tree-structured Parzen Estimators (TPE) algorithm was used to optimize model hyperparameters. The impact of climate change on IRI and rut depth was analyzed. Results showed that the proposed BNN model surpasses Artificial Neural Network (ANN), providing predictions with confidence intervals that account for uncertainty in climate data and model parameters. Compared to historical climate data, increases in IRI and rut depth were more significant when based on projected climate data. Relying only on historical climate data would underestimate pavement deterioration. Climate change appeared to have a more significant impact on rut depth than on IRI. Rut depth was particularly sensitive to climate change, increasing by more than 40%. Considering the uncertainty, rutting depth could increase by up to 85.6%. This highlights the importance of considering regional differences in climate change when developing reliable prediction models. The main contributions of this study include the quantification of uncertainty, the impact analysis of climate change and regional sensitivity analysis. It helps adapt to future climate change and supports informed decision-making in transportation infrastructure management.
{"title":"Predicting Asphalt Pavement Deterioration Under Climate Change Uncertainty Using Bayesian Neural Network","authors":"Bingyan Cui;Hao Wang","doi":"10.1109/TITS.2024.3505237","DOIUrl":"https://doi.org/10.1109/TITS.2024.3505237","url":null,"abstract":"Uncertainty in climate change poses challenges in obtaining accurate and reliable prediction models for future pavement performance. This study aimed to develop an advanced prediction model specifically for flexible pavements, incorporating uncertainty quantification through a Bayesian Neural Network (BNN). Focusing on predicting the International Roughness Index (IRI) and rut depth of asphalt pavement, BNN model was applied to different climate regions, using long-term pavement performance (LTPP) data from 1989 to 2021. The Tree-structured Parzen Estimators (TPE) algorithm was used to optimize model hyperparameters. The impact of climate change on IRI and rut depth was analyzed. Results showed that the proposed BNN model surpasses Artificial Neural Network (ANN), providing predictions with confidence intervals that account for uncertainty in climate data and model parameters. Compared to historical climate data, increases in IRI and rut depth were more significant when based on projected climate data. Relying only on historical climate data would underestimate pavement deterioration. Climate change appeared to have a more significant impact on rut depth than on IRI. Rut depth was particularly sensitive to climate change, increasing by more than 40%. Considering the uncertainty, rutting depth could increase by up to 85.6%. This highlights the importance of considering regional differences in climate change when developing reliable prediction models. The main contributions of this study include the quantification of uncertainty, the impact analysis of climate change and regional sensitivity analysis. It helps adapt to future climate change and supports informed decision-making in transportation infrastructure management.","PeriodicalId":13416,"journal":{"name":"IEEE Transactions on Intelligent Transportation Systems","volume":"26 1","pages":"785-797"},"PeriodicalIF":7.9,"publicationDate":"2024-12-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142975721","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-12-13DOI: 10.1109/TITS.2024.3516892
{"title":"2024 Index IEEE Transactions on Intelligent Transportation Systems Vol. 25","authors":"","doi":"10.1109/TITS.2024.3516892","DOIUrl":"https://doi.org/10.1109/TITS.2024.3516892","url":null,"abstract":"","PeriodicalId":13416,"journal":{"name":"IEEE Transactions on Intelligent Transportation Systems","volume":"25 12","pages":"1-312"},"PeriodicalIF":7.9,"publicationDate":"2024-12-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10798999","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142870213","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Road ponding, a prevalent traffic hazard, poses a serious threat to road safety by causing vehicles to lose control and leading to accidents ranging from minor fender benders to severe collisions. Existing technologies struggle to accurately identify road ponding due to complex road textures and variable ponding coloration influenced by reflection characteristics. To address this challenge, we propose a novel approach called Self-Attention-based Global Saliency-Enhanced Network (AGSENet) for proactive road ponding detection and traffic safety improvement. AGSENet incorporates saliency detection techniques through the Channel Saliency Information Focus (CSIF) and Spatial Saliency Information Enhancement (SSIE) modules. The CSIF module, integrated into the encoder, employs self-attention to highlight similar features by fusing spatial and channel information. The SSIE module, embedded in the decoder, refines edge features and reduces noise by leveraging correlations across different feature levels. To ensure accurate and reliable evaluation, we corrected significant mislabeling and missing annotations in the Puddle-1000 dataset. Additionally, we constructed the Foggy-Puddle and Night-Puddle datasets for road ponding detection in low-light and foggy conditions, respectively. Experimental results demonstrate that AGSENet outperforms existing methods, achieving IoU improvements of 2.03%, 0.62%, and 1.06% on the Puddle-1000, Foggy-Puddle, and Night-Puddle datasets, respectively, setting a new state-of-the-art in this field. Finally, we verified the algorithm’s reliability on edge computing devices. This work provides a valuable reference for proactive warning research in road traffic safety. The source code and datasets are placed in the https://github.com/Lyu-Dakang/AGSENet.
{"title":"AGSENet: A Robust Road Ponding Detection Method for Proactive Traffic Safety","authors":"Ronghui Zhang;Shangyu Yang;Dakang Lyu;Zihan Wang;Junzhou Chen;Yilong Ren;Bolin Gao;Zhihan Lv","doi":"10.1109/TITS.2024.3506659","DOIUrl":"https://doi.org/10.1109/TITS.2024.3506659","url":null,"abstract":"Road ponding, a prevalent traffic hazard, poses a serious threat to road safety by causing vehicles to lose control and leading to accidents ranging from minor fender benders to severe collisions. Existing technologies struggle to accurately identify road ponding due to complex road textures and variable ponding coloration influenced by reflection characteristics. To address this challenge, we propose a novel approach called Self-Attention-based Global Saliency-Enhanced Network (AGSENet) for proactive road ponding detection and traffic safety improvement. AGSENet incorporates saliency detection techniques through the Channel Saliency Information Focus (CSIF) and Spatial Saliency Information Enhancement (SSIE) modules. The CSIF module, integrated into the encoder, employs self-attention to highlight similar features by fusing spatial and channel information. The SSIE module, embedded in the decoder, refines edge features and reduces noise by leveraging correlations across different feature levels. To ensure accurate and reliable evaluation, we corrected significant mislabeling and missing annotations in the Puddle-1000 dataset. Additionally, we constructed the Foggy-Puddle and Night-Puddle datasets for road ponding detection in low-light and foggy conditions, respectively. Experimental results demonstrate that AGSENet outperforms existing methods, achieving IoU improvements of 2.03%, 0.62%, and 1.06% on the Puddle-1000, Foggy-Puddle, and Night-Puddle datasets, respectively, setting a new state-of-the-art in this field. Finally, we verified the algorithm’s reliability on edge computing devices. This work provides a valuable reference for proactive warning research in road traffic safety. The source code and datasets are placed in the <uri>https://github.com/Lyu-Dakang/AGSENet</uri>.","PeriodicalId":13416,"journal":{"name":"IEEE Transactions on Intelligent Transportation Systems","volume":"26 1","pages":"497-516"},"PeriodicalIF":7.9,"publicationDate":"2024-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142976189","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Rain deviates the distribution of rainy images and the clean, rain-free data typically used during perception model training, this kind of out-of-distribution (OOD) issue making it difficult for models to generalize effectively in rainy scenarios, leading the performance degrade of autonomous perception systems in visual tasks such as lane detection and depth estimation, posing serious safety risks. To address this issue, we propose the Ultra-Fast Deraining Plugin (UFDP), a model-efficient deraining solution specifically designed to realign the distribution of rainy images and their rain-free counterparts. UFDP not only effectively removes rain from images but also seamlessly integrates into existing visual perception models, significantly enhancing their robustness and stability under rainy conditions. Through a detailed analysis of single-image color histograms and dataset-level distribution, we demonstrate how UFDP improves the similarity between rainy and non-rainy image distributions. Additionally, qualitative and quantitative results highlight UFDP’s superiority over state-of-the-art (SOTA) methods, showing a 5.4% improvement in SSIM and 8.1% in PSNR. UFDP also excels in terms of efficiency, achieving 7 times higher FPS than the slowest method, reducing FLOPs by 53.7 times, and using 28.8 times fewer MACs, with 6.2 times fewer parameters. This makes UFDP an ideal solution for ensuring reliable performance in autonomous driving visual perception systems, particularly in challenging rainy environments.
{"title":"Ultra-Fast Deraining Plugin for Vision-Based Perception of Autonomous Driving","authors":"Jihao Li;Jincheng Hu;Pengyu Fu;Jun Yang;Jingjing Jiang;Yuanjian Zhang","doi":"10.1109/TITS.2024.3503556","DOIUrl":"https://doi.org/10.1109/TITS.2024.3503556","url":null,"abstract":"Rain deviates the distribution of rainy images and the clean, rain-free data typically used during perception model training, this kind of out-of-distribution (OOD) issue making it difficult for models to generalize effectively in rainy scenarios, leading the performance degrade of autonomous perception systems in visual tasks such as lane detection and depth estimation, posing serious safety risks. To address this issue, we propose the Ultra-Fast Deraining Plugin (UFDP), a model-efficient deraining solution specifically designed to realign the distribution of rainy images and their rain-free counterparts. UFDP not only effectively removes rain from images but also seamlessly integrates into existing visual perception models, significantly enhancing their robustness and stability under rainy conditions. Through a detailed analysis of single-image color histograms and dataset-level distribution, we demonstrate how UFDP improves the similarity between rainy and non-rainy image distributions. Additionally, qualitative and quantitative results highlight UFDP’s superiority over state-of-the-art (SOTA) methods, showing a 5.4% improvement in SSIM and 8.1% in PSNR. UFDP also excels in terms of efficiency, achieving 7 times higher FPS than the slowest method, reducing FLOPs by 53.7 times, and using 28.8 times fewer MACs, with 6.2 times fewer parameters. This makes UFDP an ideal solution for ensuring reliable performance in autonomous driving visual perception systems, particularly in challenging rainy environments.","PeriodicalId":13416,"journal":{"name":"IEEE Transactions on Intelligent Transportation Systems","volume":"26 1","pages":"1227-1240"},"PeriodicalIF":7.9,"publicationDate":"2024-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142975920","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-12-09DOI: 10.1109/TITS.2024.3506748
Kyler Nelson;Mario Harper
This paper presents POSEIDON-SAT, a novel dataset augmentation method designed to enhance the detection of fishing vessels using optical remote sensing technologies. Illegal fishing poses a significant threat to conservation and economic fishing zones, and its detection is often hindered by tactics such as the disabling or manipulation of Automatic Identification System (AIS) transponders. While convolutional neural networks (CNNs) have shown promise in ship detection from optical imagery, the fine-grained classification of fishing vessels is limited by the scarcity of detailed datasets, as these vessels are often underrepresented in existing databases. POSEIDON-SAT addresses this gap by augmenting datasets with synthesized fishing vessel instances, improving the performance of ship detection models, particularly in low-resource scenarios. This approach is tailored for use on low-power, edge computing platforms aboard small satellites, such as CubeSats, where computational resources are highly constrained. By comparing POSEIDON-SAT to traditional class-weighting techniques, we evaluate its impact on lightweight YOLO models optimized for real-time detection aboard such satellites. Our experimental results demonstrate that POSEIDON-SAT significantly improves detection accuracy while reducing false positives, making it an effective tool for enhancing the capabilities of remote sensing platforms in monitoring illegal fishing. This method holds promise for addressing the global challenge of illegal fishing through scalable, efficient satellite-based monitoring systems.
{"title":"POSEIDON-SAT: Data Enhancement for Optical Fishing Vessel Detection From Low-Cost Satellites","authors":"Kyler Nelson;Mario Harper","doi":"10.1109/TITS.2024.3506748","DOIUrl":"https://doi.org/10.1109/TITS.2024.3506748","url":null,"abstract":"This paper presents POSEIDON-SAT, a novel dataset augmentation method designed to enhance the detection of fishing vessels using optical remote sensing technologies. Illegal fishing poses a significant threat to conservation and economic fishing zones, and its detection is often hindered by tactics such as the disabling or manipulation of Automatic Identification System (AIS) transponders. While convolutional neural networks (CNNs) have shown promise in ship detection from optical imagery, the fine-grained classification of fishing vessels is limited by the scarcity of detailed datasets, as these vessels are often underrepresented in existing databases. POSEIDON-SAT addresses this gap by augmenting datasets with synthesized fishing vessel instances, improving the performance of ship detection models, particularly in low-resource scenarios. This approach is tailored for use on low-power, edge computing platforms aboard small satellites, such as CubeSats, where computational resources are highly constrained. By comparing POSEIDON-SAT to traditional class-weighting techniques, we evaluate its impact on lightweight YOLO models optimized for real-time detection aboard such satellites. Our experimental results demonstrate that POSEIDON-SAT significantly improves detection accuracy while reducing false positives, making it an effective tool for enhancing the capabilities of remote sensing platforms in monitoring illegal fishing. This method holds promise for addressing the global challenge of illegal fishing through scalable, efficient satellite-based monitoring systems.","PeriodicalId":13416,"journal":{"name":"IEEE Transactions on Intelligent Transportation Systems","volume":"26 1","pages":"1113-1122"},"PeriodicalIF":7.9,"publicationDate":"2024-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142976196","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-12-04DOI: 10.1109/TITS.2024.3491784
Ye Yue;Honggang Qi;Yongqiang Deng;Juanjuan Li;Hao Liang;Jun Miao
In recent years, with the advancement of artificial intelligence technology, autonomous driving technologies have gradually emerged. 3D object detection using point clouds has become a key in this field. Multi-frame fusion of point clouds is a promising technique to enhance 3D object detection for autonomous driving systems. However, most existing multi-frame detection methods focus primarily on utilizing vehicle-side lidar data. Infrastructure-side detection remains relatively unexplored, yet can enhance vital vehicle-road coordination capabilities. To help with this coordination, we propose an efficient multi-frame aggregation multi-scale fusion network specifically for infrastructure-side 3D object detection. First, our key innovation is a novel multi-frame feature aggregation module that effectively integrates information from multiple past point cloud frames to improve detection accuracy. This module comprises a feature pyramid network to fuse multi-scale features, as well as a cross-attention mechanism to learn semantic correlations between different frames over time. Next, we incorporate deformable attention, which reduces the computational overhead of aggregation by sampling locations. We designed Multi-frame and Multi-scale modules, thereby we named the model MAMF-Net. Finally, through extensive experiments on two infrastructure-side datasets including the V2X-Seq-SPD dataset which was released by Baidu corporation, we demonstrate that MAMF-Net delivers consistent accuracy improvements over single frame detectors such as PointPillars, PV-RCNN and TED-S, especially boosting pedestrian detection by 5%. Our approach also surpasses other multi-frame methods designed for vehicle-side point clouds such as MPPNet.
{"title":"Infrastructure-Side Point Cloud Object Detection via Multi-Frame Aggregation and Multi-Scale Fusion","authors":"Ye Yue;Honggang Qi;Yongqiang Deng;Juanjuan Li;Hao Liang;Jun Miao","doi":"10.1109/TITS.2024.3491784","DOIUrl":"https://doi.org/10.1109/TITS.2024.3491784","url":null,"abstract":"In recent years, with the advancement of artificial intelligence technology, autonomous driving technologies have gradually emerged. 3D object detection using point clouds has become a key in this field. Multi-frame fusion of point clouds is a promising technique to enhance 3D object detection for autonomous driving systems. However, most existing multi-frame detection methods focus primarily on utilizing vehicle-side lidar data. Infrastructure-side detection remains relatively unexplored, yet can enhance vital vehicle-road coordination capabilities. To help with this coordination, we propose an efficient multi-frame aggregation multi-scale fusion network specifically for infrastructure-side 3D object detection. First, our key innovation is a novel multi-frame feature aggregation module that effectively integrates information from multiple past point cloud frames to improve detection accuracy. This module comprises a feature pyramid network to fuse multi-scale features, as well as a cross-attention mechanism to learn semantic correlations between different frames over time. Next, we incorporate deformable attention, which reduces the computational overhead of aggregation by sampling locations. We designed Multi-frame and Multi-scale modules, thereby we named the model MAMF-Net. Finally, through extensive experiments on two infrastructure-side datasets including the V2X-Seq-SPD dataset which was released by Baidu corporation, we demonstrate that MAMF-Net delivers consistent accuracy improvements over single frame detectors such as PointPillars, PV-RCNN and TED-S, especially boosting pedestrian detection by 5%. Our approach also surpasses other multi-frame methods designed for vehicle-side point clouds such as MPPNet.","PeriodicalId":13416,"journal":{"name":"IEEE Transactions on Intelligent Transportation Systems","volume":"26 1","pages":"703-713"},"PeriodicalIF":7.9,"publicationDate":"2024-12-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142976074","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","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：