We propose a new method to count objects of specific categories that are significantly smaller than the ground sampling distance of a satellite image. This task is hard due to the cluttered nature of scenes where different object categories occur. Target objects can be partially occluded, vary in appearance within the same class and look alike to different categories. Since traditional object detection is infeasible due to the small size of objects with respect to the pixel size, we cast object counting as a density estimation problem. To distinguish objects of different classes, our approach combines density estimation with semantic segmentation in an end-to-end learnable convolutional neural network (CNN). Experiments show that deep semantic density estimation can robustly count objects of various classes in cluttered scenes. Experiments also suggest that we need specific CNN architectures in remote sensing instead of blindly applying existing ones from computer vision.
Guided Attention Network for Object Detection and Counting on Drones
Institute of Software Chinese Academy of Sciences(中国科学院软件研究所);University of Chinese Academy of Sciences(中国科学院大学);University at Albany, SUNY(纽约州立大学奥尔巴尼分校);JD Digits(京东数科)
Panoptic segmentation with highly imbalanced semantic labels
Max-Delbrueck-Center for Molecular Medicine in the Helmholtz Association (MDC)(亥姆霍兹联合会马克斯·德尔布吕克分子医学中心);Humboldt-Universität zu Berlin(柏林洪堡大学);Case Western Reserve University(凯斯西储大学);Lausanne University Hospital(洛桑大学医院);University of Bern(伯尔尼大学);Charité University Medicine(夏里特医科大学)
SSR-HEF: Crowd Counting with Multi-Scale Semantic Refining and Hard Example Focusing
Institute of Intelligent Machines, Hefei Institutes of Physical Science, Chinese Academy of Sciences(中国科学院合肥物质科学研究院智能机械研究所);University of Science and Technology of China(中国科学技术大学);Hefei University of Technology(合肥工业大学);University of Sydney(悉尼大学);Zhejiang Sci-Tech University(浙江理工大学)
基于密度图的人群计数通常被视为回归任务。深度学习用于学习图像内容与人群密度分布之间的映射。虽然取得了很大的成功,但一些远离摄像头的行人很难被发现。而且硬例子的数量往往更大。现有的基于简单欧氏距离算法的算法不加区分地对难样本和易样本进行优化,使得难样本的密度通常被错误地预测为较低甚至为零,从而导致较大的计数误差。为了解决这个问题,我们首先提出了Hard Example Focusing(HEF)算法,用于人群计数的回归任务。HEF算法通过衰减简单样本的贡献,使模型快速聚焦于困难样本,从而提高估计错误的困难样本的重要性。此外,人群场景中的比例变化很大,并且比例标注是劳动密集型的并且昂贵。通过提出多尺度语义提炼(SSR)策略,模型的底层可以突破深度学习的局限性,捕捉不同尺度的语义特征,以充分应对尺度变化。我们在六个基准数据集上进行了大量的实验来验证所提出的方法。结果表明,我们所提出的方法比国家的最先进的方法的优越性。此外,我们设计的模型更小,更快。
Counting Beyond Instances: A Benchmark for Group-Individual Object Counting
School of Computer Science and Technology, Huazhong University of Science and Technology(华中科技大学计算机科学与技术学院);Shanghai Artificial Intelligence Laboratory(上海人工智能实验室);University of Liverpool(利物浦大学)
CountingFruit: Language-Guided 3D Fruit Counting with Semantic Gaussian Splatting
University of Liverpool(利物浦大学);Xi'an Jiaotong-Liverpool University(西安交通大学利物浦大学);Baidu AI Cloud Group (ACG)(百度AI云集团);Hong Kong University of Science and Technology (Guangzhou)(香港科技大学(广州));University of Cambridge(剑桥大学)
Does it Really Count? Assessing Semantic Grounding in Text-Guided Class-Agnostic Counting
Institute of Information Science and Technologies of the National Research Council (ISTI-CNR)(意大利国家研究理事会信息科学与技术研究所);University of Pisa - Department of Information Engineering(比萨大学信息工程系)
Every Component Counts: Rethinking the Measure of Success for Medical Semantic Segmentation in Multi-Instance Segmentation Tasks
Karlsruhe Institute of Technology(卡尔斯鲁厄理工学院);HIDSS4Health - Helmholtz Information and Data Science School for Health(HIDSS4Health——亥姆霍兹健康信息与数据科学学院);University Hospital Essen(埃森大学医院);University of Duisburg-Essen(杜伊斯堡-埃森大学);German Cancer Consortium (DKTK)(德国癌症联盟(DKTK))
FusionCounting: Robust visible-infrared image fusion guided by crowd counting via multi-task learning
School of Information Science and Engineering, Yanshan University(燕山大学信息科学与工程学院);Fusion Intelligence Laboratory, Department of Computer Science, University of Exeter(埃克塞特大学计算机科学系融合智能实验室)
OmniCount: Multi-label Object Counting with Semantic-Geometric Priors
University of Surrey(萨里大学);CVSSP(视觉、语音与信号处理中心);Surrey Institute for People-Centred AI(萨里以人为本人工智能研究所);School of Veterinary Medicine(兽医学院);iFlyTek-Surrey Joint Research Center on AI(科大讯飞-萨里人工智能联合研究中心)
本文提出开放词汇多类别目标计数框架 OmniCount,利用预训练模型的语义-几何先验和 Segment Anything Model 实现免训练、多标签高效计数,并构建 OmniCount-191 基准验证其性能优势。
FruitNeRF: A Unified Neural Radiance Field based Fruit Counting Framework
Friedrich-Alexander-Universität Erlangen-Nürnberg-Fürth(弗里德里希-亚历山大-埃尔兰根-纽伦堡-菲尔特大学);Fraunhofer Institute for Integrated Circuits (IIS) - EZRT(弗劳恩霍夫集成电路研究所(IIS)- EZRT);University of Bamberg(班贝格大学)
T2ICount: Enhancing Cross-modal Understanding for Zero-Shot Counting
University of Nottingham(诺丁汉大学);University of St Andrews(圣安德鲁大学);City University of Hong Kong(香港城市大学);Nanyang Technology University(南洋理工大学);Harbin Institute of Technology(哈尔滨工业大学)
Boosting Quantitive and Spatial Awareness for Zero-Shot Object Counting
Northwestern Polytechnical University(西北工业大学);Institute of Artificial Intelligence (TeleAI), China Telecom(中国电信人工智能研究院);University of Science and Technology of China(中国科学技术大学);Fudan University(复旦大学)
UniCounting: Instance-Aware Proposal Consolidation for Image-Query-Free Multi-Category Counting
Shenzhen University(深圳大学);The Hong Kong University of Science and Technology (Guangzhou)(香港科技大学(广州));Hunan University of Science and Technology(湖南科技大学);Peng Cheng Laboratory(鹏城实验室);Fudan University(复旦大学)
TCFormer: A 5M-Parameter Transformer with Density-Guided Aggregation for Weakly-Supervised Crowd Counting
College of Mechanical and Electronic Engineering, Dalian Minzu University(大连民族大学机械与电子工程学院);Dalian University of Technology(大连理工大学);Postdoctoral workstation of Dalian Rijia Electronics Co., Ltd.(大连瑞嘉电子有限公司博士后工作站);School of Computer Science and Engineering, Dalian Minzu University(大连民族大学计算机科学与工程学院);R&D Department of Dalian Rijia Electronics Co., Ltd.(大连瑞嘉电子有限公司研发部)
CoHD: A Counting-Aware Hierarchical Decoding Framework for Generalized Referring Expression Segmentation
Tsinghua Shenzhen International Graduate School, Tsinghua University(清华大学深圳国际研究生院);School of EIC, Huazhong University of Science and Technology(华中科技大学电子信息与工程学院);Tencent TEG(腾讯技术工程事业群)