这是AAAI2019的文章,用GCN坐Zero-shot动作识别,个人觉得很有意思。(I Know the Relationships: Zero-Shot Action Recognition via Two-Stream Graph Convolutional Networks and Knowledge Graphs, AAAI 2019) 1、第一个有意思的地方——Three types of relationships: 因为是Zero-Shot Action Recognition(ZSAR),和ZSL...
我们考虑 zero-shot recognition 的问题:学习一个类别的视觉分类器,并且不用 training data,仅仅依赖于 类别的单词映射(the word embedding of the category)及其与其他类别的关系(its relationship to other categories),但是会提供 visual data。处理这种不熟悉的或者新颖的类别,从已有的类别中进行知识的迁移是成功的...
3.2.2 Open-set Action Localisation 上表展示了本文方法在Open-set Action Localisation任务上的实验结果。 3.3. Text-Video Retrieval 上表展示了本文方法在视频-文本检索上的实验结果。 04 总结 本文提出了用基于CLIP来做Action Recognition、Text-Video Retrieval、Action Localisation三个任务。对于前两个任务,处理...
Generalizedzero-shotactionrecognition中对可见和不可见动作类别进行单独处理来解决该问题。 本文引入了out-of-distribution检测器,该检测器确定视频特征...的基于out-of-distribution检测器的Generalizedzero-shotactionrecognition框架: OD检测器可以减少标准GZSL框架中存在的可见动作类别的固有 ...
Zero-shot Recognition via Semantic Embeddings and Knowledge Graphs 基于语义嵌入和知识图的零样本识别 1. 简介 1.1 摘要 我们考虑零样本识别的问题:为没有训练示例的类别学习一个视觉分类器,仅使用类别的词嵌入及与其他类别的关系,来提供视觉数据。处理不熟悉或新颖类别的关键是迁移从熟悉的类中获得的知识来描述不...
Zero-shot action recognition requires a strong ability to generalize from pre-training and seen classes to novel unseen classes. Similarly, continual learning aims to develop models that can generalize effectively and learn new tasks without forgetting the ones previously learned. The generalization goals...
Zero-Shot Activity Recognition with Verb Attribute Induction In this paper, we investigate large-scale zero-shot activity recognition bymodeling the visual and linguistic attributes of action verbs. For example, theverb "salute" has several properties, such as being a light movement, a socialact, ...
Transductive Zero-Shot Action Recognition by Word-Vector Embedding The number of categories for action recognition is growing rapidly and it has become increasingly hard to label sufficient training data for learning conve... X Xu,T Hospedales,S Gong - 《International Journal of Computer Vision》 ...
论文名称:A Causal View of Compositional Zero-Shot Recognition 来自NVIDIA的作品,今年NIPS的spotlight,再一次让人看到了Causality,特别是intervention在传统CV领域带来的新路子。 首先我们先来分析一下标题, A causal view代表了他与传统工作不同的地方,以因果为工具,我们可以避免NN带来严重的spurious correlation和泛化...