夜间行人检测 & 重识别
- 夜间监控场景的低照度条件会导致场景图像像素失真概率增加，进而影响目标检测/识别方法的性能。针对这个问题，实验室在国家自然科学基金的资助下提出了面向夜间监控场景的目标检测/行人重识别新课题，构建了国内外首个自建夜间显著性目标检测 & 行人检测数据集，研究了夜间场景中噪声干扰、弱纹理模糊等因素对目标特征的影响机理，揭示了夜间非均匀照明环境中目标随景深和环境光的退化规律，在F-measure、平均绝对误差MAE、ROC曲线下面积AUC等性能指标上均得到了显著提升。研究成果可为夜间安防监控、夜间自动驾驶等应用提供理论和技术支撑。
- Xin Yuan, Xin Xu, Zheng Wang, Kai Zhang, Wei Liu, and Ruimin Hu, “Searching Parameterized Retrieval & Verification Loss for Re-identification.” IEEE Journal of Selected Topics in Signal Processing, 2023. https://ieeexplore.ieee.org/abstract/document/10056933
- 别倩, 王晓, 徐新, 赵启军, 王正, 陈军, 胡瑞敏, “红外—可见光跨模态的行人检测综述.” 中国图象图形学报, 2023. http://www.cjig.cn/jig/ch/reader/view_abstract.aspx?file_no=220670&flag=1
- Xin Xu, Wei Liu, Zheng Wang, Ruimin Hu, and Qi Tian, “Towards generalizable person re-identification with a bi-stream generative model.” Pattern Recognition, 2022. https://www.sciencedirect.com/science/article/abs/pii/S0031320322004344.
- Xin Xu, Xin Yuan, Zheng Wang, Kai Zhang, and Ruimin Hu, “Rank-in-Rank Loss for Person Re-identification.” ACM Transactions on Multimedia Computing, Communications, and Applications, 2022. https://dl.acm.org/doi/10.1145/3532866
- Xiao Wang, Zheng Wang, Wu Liu, Xin Xu, Jing Chen, and Chia-Wen Lin, “Consistency-constancy bi-knowledge learning for pedestrian detection in night surveillance.” in Proceedings of the 29th ACM International Conference on Multimedia (ACM MM), Oct. 2021. https://dl.acm.org/doi/abs/10.1145/3474085.3475599
- Xiao Wang, et al., “S3D: scalable pedestrian detection via score scale surface discrimination.” IEEE Transactions on Circuits and Systems for Video Technology, 2020. https://ieeexplore.ieee.org/abstract/document/8698270
- Xiao Wang, et al., “Listen, look, and find the one: robust person search with multimodality index.” ACM Transactions on Multimedia Computing, Communications, and Applications, 2020. https://dl.acm.org/doi/abs/10.1145/3380549
- 徐新, 穆楠, 张晓龙. 一种鲁棒的夜间图像显著性对象检测模型[J]. 软件学报, 2018. http://www.jos.org.cn/1000-9825/5396.htm
低照度图像增强 & 修复
- 低照度图像增强对于高质量的图像显示及相关视觉应用非常重要，增强后的图像需要在保持其视觉自然度的同时提高可见度。针对这个需求，实验室开展了面向夜间、雾天等低照度场景的图像增强 & 修复课题，揭示了低照度场景中的颜色退化规律及其对场景中目标的影响机理，实现了失真低照度图像的色彩对比度恢复，同时有效提升了场景中目标的可见度，在主观评价指标和多个客观评价指标上均得到了显著提升。研究成果可为低照度图像处理、数字印刷、风格化黑白摄影等应用提供理论和技术支撑。
- Wanyu Wu, Wei Wang, Zheng Wang, Kui Jiang, and Xin Xu, “From Generation to Suppression: Towards Effective Irregular Glow Removal for Nighttime Visibility Enhancement.” in Proceedings of the 32th International Joint Conference on Artificial Intelligence (IJCAI), 2023. https://www.ijcai.org/proceedings/2023/170
- Xin Xu, Shiqin Wang, Zheng Wang, Xiaolong Zhang, and Ruimin Hu, “Exploring image enhancement for salient object detection in low light images.” ACM Transactions on Multimedia Computing, Communications, and Applications, 2021. https://dl.acm.org/doi/abs/10.1145/3414839
- Pengyu Xie, Xin Xu, Zheng Wang, and Toshihiko Yamasaki, “Sampling and Re-weighting: Towards Diverse Frame Aware Unsupervised Video Person Re-identification.” IEEE Transactions on Multimedia, 2022. https://ieeexplore.ieee.org/document/9807391
- Pengyu Xie, Xin Xu, Zheng Wang, and Toshihiko Yamasaki, “Unsupervised video person re-identification via noise and hard frame aware clustering.” IEEE International Conference on Multimedia and Expo (ICME), 2021. （最佳论文提名奖） https://ieeexplore.ieee.org/abstract/document/9428200
- Wei Wang, et al., “Hazy image decolorization with color contrast restoration.” IEEE Transactions on Image Processing, 2020. https://ieeexplore.ieee.org/abstract/document/8835066
- Wei Wang, et al., “Color contrast-preserving decolorization.” IEEE Transactions on Image Processing, 2018. https://ieeexplore.ieee.org/abstract/document/6215215
监控数据集构建 & 处理
- 对大规模图像数据集进行人工标注非常繁琐、耗时。一般来说，从数据采集到人工标识，可能会需要等待一个甚至数个月的时间，严重影响了研发进度。然而标注数据对于深度学习模型的训练是十分重要的。针对这个问题，实验室开展了面向大规模监控数据集的人机共生（Human-in-the-Loop）课题，揭示了现有大规模监控数据集中存在着的冗余和错误标注现象，实现了用少量的标注代价来高效训练深度学习模型，同时为人类标注员提供了自动化标注工具，将人类的反馈快速分享给机器。研究成果已成功用于标注夜间监控场景的行人检测数据集和无约束场景的重要人定位数据集，并可为相关研究的大规模数据集构建和处理提供理论和技术支撑。
- Xin Xu, Lei Liu, Xiaolong Zhang, Weili Guan, and Ruimin Hu, “Rethinking data collection for person re-identification: active redundancy reduction.” Pattern Recognition, 2021. https://www.sciencedirect.com/science/article/pii/S0031320321000145.
- Zhijing Wan, Xin Xu, Zheng Wang, Zhixiang Wang, and Ruimin Hu, “From multi-source virtual to real: effective virtual data search for vehicle re-identification.” IEEE Transactions on Intelligent Transportation Systems, 2022.
- Zhijing Wan, Xin Xu, Zheng Wang, Toshihiko Yamasaki, Xiaolong Zhang, and Ruimin Hu, “Efficient virtual data search for annotation-free vehicle reidentification.” International Journal of Intelligent Systems, 2022.（封面论文） https://onlinelibrary.wiley.com/doi/10.1002/int.22829
- Xiao Wang, et al., “Very important person localization in unconstrained conditions: a new benchmark.” in Proceedings of The Thirty-Fifth AAAI Conference on Artificial Intelligence (AAAI), 2021. https://www.aaai.org/AAAI21Papers/AAAI-727.WangX.pdf
- Xiao Wang, et al., “When pedestrian detection meets nighttime surveillance: a new benchmark.” in Conf. Rec. 2020 the Twenty-Ninth International Joint Conference on Artificial Intelligence (IJCAI), Yokohama, Tokyo, 2020. https://www.ijcai.org/Proceedings/2020/0071.pdf
大气 & 环境遥感
- 随着遥感探测技术的不断发展，海量遥感数据的激增给地球观测及地学研究带来了新的可能；人工智能方法的引入能有效提取遥感海量数据的信息，充分挖掘数据潜能。目前，实验室已将越来越多的机器学习方法应用到了监测地球变化、反演地球参数和评估地球生态环境中，包括云覆盖和云分数的评估、大气温湿度廓线的反演和垂直颗粒物分布的预测。这些研究结合了遥感数据来源多、数据量大和信息丰富的特点，充分发挥了人工智能在数据挖掘的优势，取得了良好的成果和十足的进展。目前，研究成果已成功用于卫星遥感数据的预处理，并可为天气预报、污染物监测等应用提供理论和技术支撑。
- Xin Yuan, Xin Xu, Xiao Wang, Kai Zhang, Liang Liao, Zheng Wang, and Chia‐Wen Lin, “OSAP-Loss: Efficient optimization of average precision via involving samples after positive ones towards remote sensing image retrieval.” CAAI Transactions on Intelligence Technology, 2023. https://doi.org/10.1049/cit2.12151
- Cunxing Lei, Xin Xu, Yingying Ma, Shikuan Jin, Boming Liu, and Wei Gong, “Full coverage estimation of the PM2.5 concentration across China based on an adaptive spatiotemporal approach.” IEEE Transactions on Geoscience and Remote Sensing, 2022. https://ieeexplore.ieee.org/document/9916512
- Yang Zhu, Yingying Ma, Boming Liu, Xin Xu, Shikuan Jin, and Wei Gong, “Retrieving the vertical distribution of PM2.5 mass concentration from lidar via a random forest model.” IEEE Transactions on Geoscience and Remote Sensing, 2021. https://ieeexplore.ieee.org/abstract/document/9513287
- Wenbo Yu, Xin Xu, Shikuan Jin, Yingying Ma, Boming Liu, and Wei Gong, “BP neural network retrieval for remote sensing atmospheric profile of ground-based microwave radiometer,” IEEE Geoscience and Remote Sensing Letters, 2021. https://ieeexplore.ieee.org/abstract/document/9559908/
- 邓梦娇，徐新，马盈盈，龚威，金适宽，胡瑞敏, “多层感知机结合辐射传输模型的复杂陆地表面云检测.” 电子学报, 2022. https://www.ejournal.org.cn/CN/10.12263/DZXB.20210636