arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1708.00973cs.CVcs.MM
收藏

用于视频识别的Web图像的注意转移

Attention Transfer from Web Images for Video Recognition

National University of Singapore(新加坡国立大学);University of Minnesota(明尼苏达大学)

Junnan Li, Yongkang Wong, Qi Zhao, Mohan Kankanhalli

摘要:

训练用于动作识别的基于深度学习的视频分类器需要大量标记的视频。标签过程是劳动密集型和耗时的。另一方面,大量的弱标记图像被用户上传到互联网上。为了利用丰富且高度多样化的Web图像集,一种可扩展的方法是抓取这些图像以训练基于深度学习的分类器,例如卷积神经网络(CNN)。然而,由于域转移问题,Web图像训练的深度分类器的性能在直接部署到视频时往往会下降。解决这个问题的一种方法是对视频的训练模型进行微调,但仍然需要足够数量的注释视频。在这项工作中,我们提出了一种新的方法来转移知识从图像域到视频域。所提出的方法可以适应目标域(即视频数据)与有限数量的训练数据。我们的方法将视频帧映射到一个低维特征空间,使用CNN的分类判别空间注意力图。我们设计了一种新的Siamese EnergyNet结构,通过联合优化两个损失函数来学习注意力地图上的能量函数,使得与地面真实概念相对应的注意力地图具有更高的能量。我们对两个具有挑战性的视频识别数据集(即TVHI和UCF101)进行了广泛的实验,并证明了我们所提出的方法的有效性。

英文摘要:

Training deep learning based video classifiers for action recognition requires a large amount of labeled videos. The labeling process is labor-intensive and time-consuming. On the other hand, large amount of weakly-labeled images are uploaded to the Internet by users everyday. To harness the rich and highly diverse set of Web images, a scalable approach is to crawl these images to train deep learning based classifier, such as Convolutional Neural Networks (CNN). However, due to the domain shift problem, the performance of Web images trained deep classifiers tend to degrade when directly deployed to videos. One way to address this problem is to fine-tune the trained models on videos, but sufficient amount of annotated videos are still required. In this work, we propose a novel approach to transfer knowledge from image domain to video domain. The proposed method can adapt to the target domain (i.e. video data) with limited amount of training data. Our method maps the video frames into a low-dimensional feature space using the class-discriminative spatial attention map for CNNs. We design a novel Siamese EnergyNet structure to learn energy functions on the attention maps by jointly optimizing two loss functions, such that the attention map corresponding to a ground truth concept would have higher energy. We conduct extensive experiments on two challenging video recognition datasets (i.e. TVHI and UCF101), and demonstrate the efficacy of our proposed method.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑