Text-to-image generation models represent the next step of evolution in image synthesis, offering a natural way to achieve flexible yet fine-grained control over the result. One emerging area of research is the fast adaptation of large text-to-image models to smaller datasets or new visual concepts. However, many efficient methods of adaptation have a long training time, which limits their practical applications, slows down experiments, and spends excessive GPU resources. In this work, we study the training dynamics of popular text-to-image personalization methods (such as Textual Inversion or DreamBooth), aiming to speed them up. We observe that most concepts are learned at early stages and do not improve in quality later, but standard training convergence metrics fail to indicate that. Instead, we propose a simple drop-in early stopping criterion that only requires computing the regular training objective on a fixed set of inputs for all training iterations. Our experiments on Stable Diffusion for 48 different concepts and three personalization methods demonstrate the competitive performance of our approach, which makes adaptation up to 8 times faster with no significant drops in quality.
UASTrack: A Unified Adaptive Selection Framework with Modality-Customization in Single Object Tracking
School of Artificial Intelligence and Computer Science, Jiangnan University（江南大学人工智能与计算机科学学院）；Centre for Vision, Speech and Signal Processing, University of Surrey（萨里大学视觉、语音和信号处理中心）
Peng Cheng Laboratory（鹏城实验室）；Chongqing Normal University（重庆师范大学）；Harbin Institute of Technology, Shenzhen（哈尔滨工业大学（深圳））；Dalian University of Technology（大连理工大学）；Google Research（谷歌研究院）；University of California, Merced（加州大学默塞德分校）；Yonsei University（延世大学）
Tracking Tensor Subspaces with Informative Random Sampling for Real-Time MR Imaging
Stanford University（斯坦福大学）；University of Minnesota（明尼苏达大学）；Center for Magnetic Resonance Research（磁共振研究中心）；University of Minnesota Medical School（明尼苏达大学医学院）
BioTracker: An Open-Source Computer Vision Framework for Visual Animal Tracking
Freie Universität Berlin（柏林自由大学）；University of Applied Science Kempten（肯普滕应用科学大学）；Humboldt-Universität zu Berlin（柏林洪堡大学）；Leibniz-Institute of Freshwater Ecology and Inland Fisheries（莱布尼茨淡水生态与内陆渔业研究所）
Benchmarking YOLOv5 and YOLOv7 models with DeepSORT for droplet tracking applications
Fondazione Istituto Italiano di Tecnologia (IIT)（意大利技术研究院）；Istituto Nanoscienze-CNR（意大利国家研究委员会纳米科学研究所）；Scuola Normale Superiore（比萨高等师范学校）；Istituto per le Applicazioni del Calcolo del Consiglio Nazionale delle Ricerche（意大利国家研究委员会计算应用研究所）；Università degli Studi Roma tre（罗马第三大学）；Università di Pisa（比萨大学）
在微流体中跟踪液滴是一项具有挑战性的任务。在选择工具来分析一般微流体视频以推断物理量方面出现了困难。最先进的对象检测器算法You Only Look Once（YOLO）和对象跟踪算法Simple Online and Realtime Tracking with a Deep Association Metric（DeepSORT）可定制用于液滴识别和跟踪。定制包括训练YOLO和DeepSORT网络来识别和跟踪感兴趣的对象。我们训练了几个YOLOv5和YOLOv7模型以及DeepSORT网络，用于从微流体实验视频中识别和跟踪液滴。我们比较了液滴跟踪应用程序与YOLOv5和YOLOv7在训练时间和时间方面的性能，以分析不同硬件配置的给定视频。尽管最新的YOLOv7速度提高了10%，但由于DeepSORT算法带来的额外的显著液滴跟踪成本，只有更轻的YOLO模型在RTX 3070 Ti GPU机器上才能实现实时跟踪。这项工作是使用DeepSORT的YOLOv5和YOLOv7网络在微流体液滴自定义数据集的训练时间和推理时间方面的基准研究。
QDTrack: Quasi-Dense Similarity Learning for Appearance-Only Multiple Object Tracking
ETH Zurich（苏黎世联邦理工学院）；Shanghai AI Laboratory（上海人工智能实验室）；Massachusetts Institute of Technology（麻省理工学院）；Stanford University（斯坦福大学）；UC Berkeley（加州大学伯克利分校）
提出类稠密相似性学习并构建 QDTrack，通过稠密区域对比学习和最近邻关联实现仅外观多目标跟踪，在 BDD100K MOT 上达到新最优且开销极小。
Multiview 2D/3D Rigid Registration via a Point-Of-Interest Network for Tracking and Triangulation ($\text{POINT}^2$)
University of Rochester（罗切斯特大学）；University of Maryland, College Park（马里兰大学帕克分校）；Rutgers University（罗格斯大学）；Z2W Corporation（Z2W公司）；Chinese Academy of Sciences（中国科学院）；Peng Cheng Laboratory（鹏城实验室）
Estimation of Orofacial Kinematics in Parkinson's Disease: Comparison of 2D and 3D Markerless Systems for Motion Tracking
KITE — Toronto Rehabilitation Institute — University Health Network（凯特大西洋康复研究所——多伦多康复研究所——大学医疗网络）；University of Toronto（多伦多大学）；Sunnybrook Research Institute（森尼布鲁克研究所）
Infrared UAV Target Tracking with Dynamic Feature Refinement and Global Contextual Attention Knowledge Distillation
School of Computer Science and Technology, Xidian University（西安电子科技大学计算机科学与技术学院）；Xi’an Modern Control Technology Research Institute（西安现代控制技术研究所）；School of Artificial Intelligence and Automation, Huazhong University of Science and Technology（华中科技大学人工智能与自动化学院）
IRIS: Intent Resolution via Inference-time Saccades for Open-Ended VQA in Large Vision-Language Models
Department of Computer Science, UC Santa Barbara（计算机科学系，加州大学圣芭芭拉分校）；Department of Psychological & Brain Sciences, UC Santa Barbara（心理学与脑科学系，加州大学圣芭芭拉分校）；DEVCOM Army Research Laboratory（国防部陆军研究实验室）
