The ability to handle large scale variations is crucial for many real world visual tasks. A straightforward approach for handling scale in a deep network is to process an image at several scales simultaneously in a set of scale channels. Scale invariance can then, in principle, be achieved by using weight sharing between the scale channels together with max or average pooling over the outputs from the scale channels. The ability of such scale channel networks to generalise to scales not present in the training set over significant scale ranges has, however, not previously been explored.
In this paper, we present a systematic study of this methodology by implementing different types of scale channel networks and evaluating their ability to generalise to previously unseen scales. We develop a formalism for analysing the covariance and invariance properties of scale channel networks, and explore how different design choices, unique to scaling transformations, affect the overall performance of scale channel networks. We first show that two previously proposed scale channel network designs do not generalise well to scales not present in the training set. We explain theoretically and demonstrate experimentally why generalisation fails in these cases.
We then propose a new type of foveated scale channel architecture}, where the scale channels process increasingly larger parts of the image with decreasing resolution. This new type of scale channel network is shown to generalise extremely well, provided sufficient image resolution and the absence of boundary effects. Our proposed FovMax and FovAvg networks perform almost identically over a scale range of 8, also when training on single scale training data, and do also give improved performance when learning from datasets with large scale variations in the small sample regime.
MIDeepSeg: Minimally Interactive Segmentation of Unseen Objects from Medical Images Using Deep Learning
University of Electronic Science and Technology of China(电子科技大学);SenseTime Research(商汤科技研究院);Shanghai Jiao Tong University(上海交通大学);University Hospitals Leuven(鲁汶大学医院);University College London(伦敦大学学院);King’s College London(伦敦国王学院)
Generative Video Diffusion for Unseen Novel Semantic Video Moment Retrieval
Queen Mary University of London(伦敦玛丽女王大学);Sony AI(索尼人工智能公司);Adobe Research(奥多比研究院);WICT, Peking University(北京大学信息科学技术学院);State Key Laboratory of General Artificial Intelligence, Peking University(北京大学通用人工智能国家重点实验室)
Robust Distribution Alignment for Industrial Anomaly Detection under Distribution Shift
Institute for Infocomm Research, A*STAR(新加坡科研局信息通信研究院);Nanyang Technological University(南洋理工大学);South China University of Technology(华南理工大学);National University of Singapore(新加坡国立大学)
Evaluation of Medical Vision Language Models HuluMed and MedGemma, and general purpose chatbots Gemma 3, ChatGPT Plus, and Claude Pro on real previously unseen wound images
Department of Computer Science, New Jersey Institute of Technology(新泽西理工学院计算机科学系);Vascular and Endovascular Surgery, Robert Wood Johnson Hospital(罗伯特·伍德·约翰逊医院血管外科);Department of Data Science, New Jersey Institute of Technology(新泽西理工学院数据科学系)
本研究评估了六种视觉语言模型在慢性伤口分析任务上的表现,发现通用模型 ChatGPT 和 Claude 显著优于医学专用模型,表明广泛的多模态推理能力比领域知识更重要。