arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 1109.3737cs.AI
收藏

学习如何使用深度架构进行图像跟踪

Learning where to Attend with Deep Architectures for Image Tracking

University of British Columbia(不列颠哥伦比亚大学);University of Verona(维罗纳大学);University of Sherbrooke(舍布鲁克大学)

Misha Denil, Loris Bazzani, Hugo Larochelle, Nando de Freitas

摘要:

我们讨论了一个注意力模型,同时对象跟踪和识别,是由凝视数据驱动。受感知理论的启发,该模型由两个相互作用的通路组成:身份和控制,旨在反映神经科学模型中的什么和哪里通路。身份路径模型对象的外观和执行分类使用深(因子)限制玻尔兹曼机。在每一个时间点上,观察结果都是由中央凹图像组成的,分辨率朝着注视的边缘逐渐衰减。控制路径模拟被关注对象的位置、方向、规模和速度。这些状态的后验分布估计与粒子滤波。在控制通路的更深处,我们遇到了一种注意力机制,它学会选择凝视,以最大限度地减少跟踪的不确定性。与我们以前的工作不同,我们引入了凝视选择策略,该策略在部分信息的存在下和连续的动作空间上操作。我们表明,一个简单的扩展现有的方法的部分信息设置的结果在性能不佳,我们提出了一种替代方法的基础上建模的奖励表面作为高斯过程。这种方法在部分信息的存在下提供了良好的性能,并允许我们将动作空间从一个小的离散的固定点集扩展到一个连续的域。

英文摘要:

We discuss an attentional model for simultaneous object tracking and recognition that is driven by gaze data. Motivated by theories of perception, the model consists of two interacting pathways: identity and control, intended to mirror the what and where pathways in neuroscience models. The identity pathway models object appearance and performs classification using deep (factored)-Restricted Boltzmann Machines. At each point in time the observations consist of foveated images, with decaying resolution toward the periphery of the gaze. The control pathway models the location, orientation, scale and speed of the attended object. The posterior distribution of these states is estimated with particle filtering. Deeper in the control pathway, we encounter an attentional mechanism that learns to select gazes so as to minimize tracking uncertainty. Unlike in our previous work, we introduce gaze selection strategies which operate in the presence of partial information and on a continuous action space. We show that a straightforward extension of the existing approach to the partial information setting results in poor performance, and we propose an alternative method based on modeling the reward surface as a Gaussian Process. This approach gives good performance in the presence of partial information and allows us to expand the action space from a small, discrete set of fixation points to a continuous domain.

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。

↑