杜松林，副教授，博士生导师，入选东南大学“至善青年学者”A层次、东南大学“小米青年学者”、江苏省科协青年科技人才托举工程、深圳市优秀科技创新人才培养项目。主持国家自然科学基金项目、江苏省基础研究计划（自然科学基金）项目、广东省基础与应用基础研究基金项目、深圳市基础研究专项（自然科学基金）项目、南京市留学人员科技创新择优资助项目、东南大学优秀青年教师科学研究资助项目、中国航天科工集团***项目、三星电子技术研发项目等，作为科研骨干参与国家重点研发计划项目、江苏省前沿技术研发计划项目、国家电网公司总部科技项目、日本学术振兴会基础研究项目、早稻田大学特别研究课题等。在IEEE TPAMI、IEEE TIP、IEEE TCSVT、IEEE TMM、IEEE TII、IEEE TASE、IEEE TITS、Pattern Recognition和CVPR、ICCV、ECCV、AAAI等学术期刊/会议发表论文八十余篇，出版英文专著Human Pose Analysis (Singapore: Springer Nature, 2024)。研究成果获ISPACS2017国际会议“最佳论文奖”、ICISIP2018国际会议“最佳报告奖”、ISIPS2018国际会议“杰出论文奖”、AIVR2019国际会议计算机视觉分会场“最佳报告奖”等学术奖项。主讲本科生“人工智能”和“模式识别与机器学习”课程、研究生“人工智能及其应用”课程，被评为东南大学本科生优秀导师。担任国家人工智能学院导师、国家自然科学基金函评专家、教育部学位中心研究生学位论文评审专家、国家留学基金评审专家、广东省科技咨询专家、南京市科技咨询专家、航空科学基金评审专家、三星电子中国研发中心技术顾问，以及IEEE TPAMI、IEEE TIP、IEEE TII、IEEE TCSVT、IEEE TASE、IEEE TIM、IEEE TGRS、IEEE/CAA JAS、IEEE TCE、Pattern Recognition和CVPR、ICCV、AAAI等学术期刊/会议审稿人。兼任电气与电子工程师协会（IEEE）高级会员、中国图象图形学学会机器视觉专委会委员、中国图象图形学学会视觉大数据专委会委员、中国图象图形学学会多媒体专委会委员、中国自动化学会模式识别与机器智能专委会委员、中国自动化学会机器人专委会委员、江苏省自动化学会学术工委会秘书长。
招生计划：
每年招收博士研究生1~2名、硕士研究生3~4名，研究方向主要是三维机器视觉与具身空间智能，详见songlin.ac.cn。
专著:
☐ Songlin Du and Takeshi Ikenaga, Human Pose Analysis, Springer Nature, 2025.
论文:
☐ Towards Free-form Local Feature Matching
Xiaoyong Lu, Songlin Du*, Yaping Yan, Xiaobo Lu, Takeshi Ikenaga
IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 48, no. 2, pp. 1470-1484, 2026.
☐ JamMa: Ultra-lightweight Local Feature Matching with Joint Mamba
Xiaoyong Lu and Songlin Du*
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 14934-14943, 2025.
☐ A Unified fNIRS Classification Framework Informed by Local Brain Activation Patterns
Zenghui Wang and Songlin Du*
IEEE Transactions on Industrial Informatics, 2025. DOI: 10.1109/TII.2025.3632147
☐ MambaMatch: Establishing Reliable Correspondences via Multi-Scale State Space Model
Xiangyang Miao, Shunxing Chen, Xinyu Liu, Shiping Wang, Songlin Du, Lianghua He, Guobao Xiao*
IEEE Transactions on Image Processing, vol. 34, pp. 7528-7541, 2025.
☐ Spatially-Aware Adaptive Diffusion: Unifying Low-Resolution Image Fusion and Super-Resolution
Jiajia Fu, Zhenni Yu, Haosheng Chen, Songlin Du, Changcai Yang, Lianghua He, Guobao Xiao*
IEEE Transactions on Circuits and Systems for Video Technology, 2025.
☐ Tex2Sem: Learning from Textures to Semantics for Robust Semantic Correspondence
Zenghui Wang, Songlin Du*, Yaping Yan, Guobao Xiao, Xiaobo Lu
IEEE Transactions on Circuits and Systems for Video Technology, vol. 35, no. 11, pp. 10875-10890, 2025.
☐ Topology Learning for Two-view Correspondence Filtering
Ziwei Shi, Xiangyang Miao, Guobao Xiao*, Songlin Du, Zheng Wang, Heng Tao Shen
IEEE Transactions on Multimedia, vol. 27, pp. 7533-7545, 2025.
☐ Skeleton-Aware Representation of Spatio-Temporal Kinematics for 3D Human Motion Prediction
Songlin Du*, Zhihan Zhuang, Zenghui Wang, Yuan Li, Takeshi Ikenaga
IEEE Transactions on Automation Science and Engineering, vol. 22, pp. 12884-12898, 2025.
☐ Bi-Pose: Bidirectional 2D-3D Transformation for Human Pose Estimation from a Monocular Camera
Songlin Du*, Hao Wang, Zhiwei Yuan, Takeshi Ikenaga
IEEE Transactions on Automation Science and Engineering, vol. 21, no. 3, pp. 3483-3496, 2024.
☐ ContextMatcher: Detector-Free Feature Matching with Cross-Modality Context
Dongyue Li and Songlin Du*
IEEE Transactions on Circuits and Systems for Video Technology, vol. 34, no. 9, pp. 7922-7934, 2024.
☐ Raising the Ceiling: Conflict-Free Local Feature Matching with Dynamic View Switching
Xiaoyong Lu and Songlin Du*
European Conference on Computer Vision (ECCV), pp. 256-273, 2024.
☐ Toward Better Generalization: Shape Feature-Enhanced Fastener Defect Detection with Diffusion Model
Shixiang Su, Songlin Du, Dezhou Wang, Xiaobo Lu*
IEEE Transactions on Instrumentation and Measurement, vol. 73, pp. 1-13, 2024.
☐ Straight-Line Detection within 1 Millisecond per Frame for Ultra-High-Speed Industrial Automation
Songlin Du*, Ziwei Dong, Yuan Li, Takeshi Ikenaga
IEEE Transactions on Industrial Informatics, vol. 19, no. 4, pp. 5965-5975, 2023.
☐ RFS-Net: Railway Track Fastener Segmentation Network with Shape Guidance
Shixiang Su, Songlin Du, Xuan Wei, Xiaobo Lu*
IEEE Transactions on Circuits and Systems for Video Technology, vol. 33, no. 3, pp. 1398-1412, 2023.
☐ Scene-aware Feature Matching
Xiaoyong Lu, Yaping Yan, Tong Wei, Songlin Du*
IEEE/CVF International Conference on Computer Vision (ICCV), pp. 3681-3690, 2023.
☐ ParaFormer: Parallel Attention Transformer for Efficient Feature Match
Xiaoyong Lu, Yaping Yan, Bin Kang, Songlin Du*
AAAI Conference on Artificial Intelligence (AAAI), pp. 1853-1860, 2023.
☐ Automatic Foreground Detection at 784 FPS for Ultra-High-Speed Human-Machine Interactions
Songlin Du*, Peikun Cai, Tingting Hu, Takeshi Ikenaga
IEEE Transactions on Automation Science and Engineering, vol. 19, no. 4, pp. 3587-3600, 2022.
☐ Subpixel Displacement Measurement at 784 FPS: From Algorithm to Hardware System
Songlin Du*, Kaidong Gu, Takeshi Ikenaga
IEEE Transactions on Instrumentation and Measurement, vol. 71, pp. 5008310:1-5008310:10, 2022.
☐ Geometric Constraint and Image Inpainting Based Railway Track Fastener Sample Generation for Improving Defect Inspection
Shixiang Su, Songlin Du, Xiaobo Lu*
IEEE Transactions on Intelligent Transportation Systems, vol. 23, no. 12, pp. 23883-23895, 2022.
☐ Highly-Parallel Hardwired Deep Convolutional Neural Network for 1-ms Dual-Hand Tracking
Peiqi Zhang, Tingting Hu, Dingli Luo, Songlin Du, Takeshi Ikenaga*
IEEE Transactions on Circuits and Systems for Video Technology, vol. 32, no. 12, pp. 8192-8203, 2022.
☐ Parallel Consensus Transformer for Local Feature Matching
Xiaoyong Lu, Yuhan Chen, Bin Kang, Songlin Du*
Pattern Recognition, vol. 173, pp. 112905:1-112905:11, 2026.
☐ Robust Unsupervised Visual Tracking Via Image-to-Video Identity Knowledge Transferring
Bin Kang, Zongyu Wang, Dong Liang, Tianyu Ding, Songlin Du*
Pattern Recognition, vol. 171, pp. 112109:1-112109:10, 2026.
☐ HSENet: Hierarchical Semantic-Enriched Network for Multi-Modal Image Fusion
Xinyu Liu, Rui Ming, Songlin Du, Lianghua He, Haibo Luo*, Guobao Xiao
Pattern Recognition, vol. 170, pp. 112043:1-112043:13, 2026.
☐ AnatPose: Bidirectionally Learning Anatomy-Aware Heatmaps for Human Pose Estimation
Songlin Du*, Zhiwen Zhang, Takeshi Ikenaga
Pattern Recognition, vol. 155, pp. 110654:1-110654:14, 2024.
☐ Kinematics-Aware Spatial-Temporal Feature Transform for 3D Human Pose Estimation
Songlin Du*, Zhiwei Yuan, Takeshi Ikenaga
Pattern Recognition, vol. 150, pp. 110316:1-110316:10, 2024.
☐ JoyPose: Jointly Learning Evolutionary Data Augmentation and Anatomy-Aware Global-Local Representation for 3D Human Pose Estimation
Songlin Du*, Zhiwei Yuan, Takeshi Ikenaga
Pattern Recognition, vol. 147, pp. 110116:1-110116:12, 2024.
☐ Fine-grained Recognition via Submodular Optimization Regulated Progressive Training
Bin Kang, Songlin Du*, Dong Liang, Fan Wu, Xin Li
Pattern Recognition, vol. 156, pp. 110849:1-110849:9, 2024.
发明专利：
☐ 基于联合数据增强和网络训练模型的3D人体姿态估计方法；专利号：ZL202110569187.7；授权公告日：2022年11月01日
☐ 一种基于并联交互架构模型实现多维度特征融合的目标检测方法；专利号：ZL 2022 1 1420718.7；授权公告日：2025年08月15日
☐ 基于注意力机制和图像特征融合的3D点云目标检测方法；专利号：ZL 2022 1 0659670.9；授权公告日：2024年11月26日
☐ 联合混合尺度特征描述子与邻居一致性的图像匹配方法；专利号：ZL 2022 1 1500472.4；授权公告日：2025年10月17日
☐ 一种基于动态图卷积神经网络的点云分析方法；专利号：ZL 2021 1 0616319.7；授权公告日：2024年11月26日
☐ 一种基于并行注意力机制的高效特征匹配方法；专利号：ZL 2023 1 0146914.8；授权公告日：2025年10月14日
☐ 一种联合稀疏和稠密邻域一致性的图像匹配方法；专利号：ZL 2022 1 1382001.8；授权公告日：2025年10月21日；
☐ 一种融合局部和全局特征的三维人体姿态估计方法；专利号：ZL 2022 1 0142459.X；授权公告日：2024年11月26日
☐ 一种基于注意力机制和邻域一致性的特征匹配方法；专利号：ZL 2022 1 0438479.1；授权公告日：2024年11月26日
☐ 一种融合点特征和网格特征的点云目标检测方法；专利号：ZL 2021 1 1398672.9；授权公告日：2024年08月23日
IEEE Senior Member
中国图象图形学学会（CSIG）机器视觉专委会委员
中国图象图形学学会（CSIG）视觉大数据专委会委员
中国图象图形学学会（CSIG）多媒体专委会委员
中国自动化学会（CAA）模式识别与机器智能专委会委员
中国自动化学会（CAA）机器人专委会委员
江苏省自动化学会（JSAA）学术工委会秘书长
本科生课程：模式识别与机器学习 (课程思政校级示范课、“AI+”建设课程)，自动化/机器人工程专业
研究生课程：人工智能及其应用 (课程思政校级示范课、“AI+”建设课程)，电子信息类专业
在计算机视觉领域，常见的相机系统通常采用二维传感器，这意味着它们捕捉到的图像是以二维数据形式呈现的。然而诸多场景下的智能体需要感知其周围的真实三维环境并确定自身的位姿，如自动驾驶、机器人导航、增强现实等，因此从二维图像恢复三维空间结构是一类重要且有挑战性的问题。欢迎同学们围绕该类问题研究和交流，例如跨视角图像匹配、运动恢复结构、相机位姿估计、目标位姿估计、人体等非刚性姿态估计等。
专著：
☐ Songlin Du and Takeshi Ikenaga, Human Pose Analysis, Springer Nature, 2025.
论文：
☐ Towards Free-form Local Feature Matching
Xiaoyong Lu, Songlin Du*, Yaping Yan, Xiaobo Lu, Takeshi Ikenaga
IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 48, no. 2, pp. 1470-1484, 2026.
☐ JamMa: Ultra-lightweight Local Feature Matching with Joint Mamba
Xiaoyong Lu and Songlin Du*
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 14934-14943, 2025.
☐ A Unified fNIRS Classification Framework Informed by Local Brain Activation Patterns
Zenghui Wang and Songlin Du*
IEEE Transactions on Industrial Informatics, 2025. DOI: 10.1109/TII.2025.3632147
☐ MambaMatch: Establishing Reliable Correspondences via Multi-Scale State Space Model
Xiangyang Miao, Shunxing Chen, Xinyu Liu, Shiping Wang, Songlin Du, Lianghua He, Guobao Xiao*
IEEE Transactions on Image Processing, vol. 34, pp. 7528-7541, 2025.
☐ Spatially-Aware Adaptive Diffusion: Unifying Low-Resolution Image Fusion and Super-Resolution
Jiajia Fu, Zhenni Yu, Haosheng Chen, Songlin Du, Changcai Yang, Lianghua He, Guobao Xiao*
IEEE Transactions on Circuits and Systems for Video Technology, 2025.
☐ Tex2Sem: Learning from Textures to Semantics for Robust Semantic Correspondence
Zenghui Wang, Songlin Du*, Yaping Yan, Guobao Xiao, Xiaobo Lu
IEEE Transactions on Circuits and Systems for Video Technology, vol. 35, no. 11, pp. 10875-10890, 2025.
☐ Topology Learning for Two-view Correspondence Filtering
Ziwei Shi, Xiangyang Miao, Guobao Xiao*, Songlin Du, Zheng Wang, Heng Tao Shen
IEEE Transactions on Multimedia, vol. 27, pp. 7533-7545, 2025.
☐ Skeleton-Aware Representation of Spatio-Temporal Kinematics for 3D Human Motion Prediction
Songlin Du*, Zhihan Zhuang, Zenghui Wang, Yuan Li, Takeshi Ikenaga
IEEE Transactions on Automation Science and Engineering, vol. 22, pp. 12884-12898, 2025.
☐ Bi-Pose: Bidirectional 2D-3D Transformation for Human Pose Estimation from a Monocular Camera
Songlin Du*, Hao Wang, Zhiwei Yuan, Takeshi Ikenaga
IEEE Transactions on Automation Science and Engineering, vol. 21, no. 3, pp. 3483-3496, 2024.
☐ ContextMatcher: Detector-Free Feature Matching with Cross-Modality Context
Dongyue Li and Songlin Du*
IEEE Transactions on Circuits and Systems for Video Technology, vol. 34, no. 9, pp. 7922-7934, 2024.
☐ Raising the Ceiling: Conflict-Free Local Feature Matching with Dynamic View Switching
Xiaoyong Lu and Songlin Du*
European Conference on Computer Vision (ECCV), pp. 256-273, 2024.
☐ Toward Better Generalization: Shape Feature-Enhanced Fastener Defect Detection with Diffusion Model
Shixiang Su, Songlin Du, Dezhou Wang, Xiaobo Lu*
IEEE Transactions on Instrumentation and Measurement, vol. 73, pp. 1-13, 2024.
☐ Straight-Line Detection within 1 Millisecond per Frame for Ultra-High-Speed Industrial Automation
Songlin Du*, Ziwei Dong, Yuan Li, Takeshi Ikenaga
IEEE Transactions on Industrial Informatics, vol. 19, no. 4, pp. 5965-5975, 2023.
☐ RFS-Net: Railway Track Fastener Segmentation Network with Shape Guidance
Shixiang Su, Songlin Du, Xuan Wei, Xiaobo Lu*
IEEE Transactions on Circuits and Systems for Video Technology, vol. 33, no. 3, pp. 1398-1412, 2023.
☐ Scene-aware Feature Matching
Xiaoyong Lu, Yaping Yan, Tong Wei, Songlin Du*
IEEE/CVF International Conference on Computer Vision (ICCV), pp. 3681-3690, 2023.
☐ ParaFormer: Parallel Attention Transformer for Efficient Feature Match
Xiaoyong Lu, Yaping Yan, Bin Kang, Songlin Du*
AAAI Conference on Artificial Intelligence (AAAI), pp. 1853-1860, 2023.
☐ Automatic Foreground Detection at 784 FPS for Ultra-High-Speed Human-Machine Interactions
Songlin Du*, Peikun Cai, Tingting Hu, Takeshi Ikenaga
IEEE Transactions on Automation Science and Engineering, vol. 19, no. 4, pp. 3587-3600, 2022.
☐ Subpixel Displacement Measurement at 784 FPS: From Algorithm to Hardware System
Songlin Du*, Kaidong Gu, Takeshi Ikenaga
IEEE Transactions on Instrumentation and Measurement, vol. 71, pp. 5008310:1-5008310:10, 2022.
☐ Geometric Constraint and Image Inpainting Based Railway Track Fastener Sample Generation for Improving Defect Inspection
Shixiang Su, Songlin Du, Xiaobo Lu*
IEEE Transactions on Intelligent Transportation Systems, vol. 23, no. 12, pp. 23883-23895, 2022.
☐ Highly-Parallel Hardwired Deep Convolutional Neural Network for 1-ms Dual-Hand Tracking
Peiqi Zhang, Tingting Hu, Dingli Luo, Songlin Du, Takeshi Ikenaga*
IEEE Transactions on Circuits and Systems for Video Technology, vol. 32, no. 12, pp. 8192-8203, 2022.
☐ Parallel Consensus Transformer for Local Feature Matching
Xiaoyong Lu, Yuhan Chen, Bin Kang, Songlin Du*
Pattern Recognition, vol. 173, pp. 112905:1-112905:11, 2026.
☐ Robust Unsupervised Visual Tracking Via Image-to-Video Identity Knowledge Transferring
Bin Kang, Zongyu Wang, Dong Liang, Tianyu Ding, Songlin Du*
Pattern Recognition, vol. 171, pp. 112109:1-112109:10, 2026.
☐ HSENet: Hierarchical Semantic-Enriched Network for Multi-Modal Image Fusion
Xinyu Liu, Rui Ming, Songlin Du, Lianghua He, Haibo Luo*, Guobao Xiao
Pattern Recognition, vol. 170, pp. 112043:1-112043:13, 2026.
☐ AnatPose: Bidirectionally Learning Anatomy-Aware Heatmaps for Human Pose Estimation
Songlin Du*, Zhiwen Zhang, Takeshi Ikenaga
Pattern Recognition, vol. 155, pp. 110654:1-110654:14, 2024.
☐ Kinematics-Aware Spatial-Temporal Feature Transform for 3D Human Pose Estimation
Songlin Du*, Zhiwei Yuan, Takeshi Ikenaga
Pattern Recognition, vol. 150, pp. 110316:1-110316:10, 2024.
☐ JoyPose: Jointly Learning Evolutionary Data Augmentation and Anatomy-Aware Global-Local Representation for 3D Human Pose Estimation
Songlin Du*, Zhiwei Yuan, Takeshi Ikenaga
Pattern Recognition, vol. 147, pp. 110116:1-110116:12, 2024.
☐ Fine-grained Recognition via Submodular Optimization Regulated Progressive Training
Bin Kang, Songlin Du*, Dong Liang, Fan Wu, Xin Li
Pattern Recognition, vol. 156, pp. 110849:1-110849:9, 2024.
发明专利：
☐ 基于联合数据增强和网络训练模型的3D人体姿态估计方法；专利号：ZL202110569187.7；授权公告日：2022年11月01日
☐ 一种基于并联交互架构模型实现多维度特征融合的目标检测方法；专利号：ZL 2022 1 1420718.7；授权公告日：2025年08月15日
☐ 基于注意力机制和图像特征融合的3D点云目标检测方法；专利号：ZL 2022 1 0659670.9；授权公告日：2024年11月26日
☐ 联合混合尺度特征描述子与邻居一致性的图像匹配方法；专利号：ZL 2022 1 1500472.4；授权公告日：2025年10月17日
☐ 一种基于动态图卷积神经网络的点云分析方法；专利号：ZL 2021 1 0616319.7；授权公告日：2024年11月26日
☐ 一种基于并行注意力机制的高效特征匹配方法；专利号：ZL 2023 1 0146914.8；授权公告日：2025年10月14日
☐ 一种联合稀疏和稠密邻域一致性的图像匹配方法；专利号：ZL 2022 1 1382001.8；授权公告日：2025年10月21日；
☐ 一种融合局部和全局特征的三维人体姿态估计方法；专利号：ZL 2022 1 0142459.X；授权公告日：2024年11月26日
☐ 一种基于注意力机制和邻域一致性的特征匹配方法；专利号：ZL 2022 1 0438479.1；授权公告日：2024年11月26日
☐ 一种融合点特征和网格特征的点云目标检测方法；专利号：ZL 2021 1 1398672.9；授权公告日：2024年08月23日
联系我们：
江苏省南京市四牌楼2号中心楼 东南大学自动化学院
电话：025-83792724
邮箱：jupiter@seu.edu.cn
