A Comprehensive Analysis of Deep Regression
A Comprehensive Analysis of Deep Regression
复制标题
DOI：
10.1109/tpami.2019.2910523
复制
发表时间：
2020-09-01
影响因子：
23.6
通讯作者：
Horaud, Radu
中科院分区：
文献类型：
--
作者：
Lathuiliere, Stephane;Mesejo, Pablo;Horaud, Radu
深度学习给数据科学带来了革命性的变化，最近它的受欢迎程度呈指数级增长，使用深度网络的论文数量也是如此。视觉任务，如人体姿势估计，也没有逃脱这一趋势。有大量的深度模型，在这些模型中，网络架构或数据预处理中的微小变化，加上优化过程的随机性质，会产生显著不同的结果，使得筛选显著优于其他方法的方法变得极其困难。这种情况促使本研究对香草深度回归，即具有线性回归顶层的卷积神经网络进行了系统的评估和统计分析。这是首次对深度回归技术进行全面分析。我们在四个视觉问题上进行了实验，并报告了中位数性能的可信区间以及结果的统计意义(如果有的话)。令人惊讶的是，由于不同的数据预处理过程导致的可变性通常超过了由于网络架构中的修改而产生的可变性。我们的结果强化了这一假设，根据这一假设，一般而言，充分调整的通用网络(例如，VGG-16或ResNet-50)可以产生接近最先进水平的结果，而不必求助于更复杂和特别的回归模型。
Deep learning revolutionized data science, and recently its popularity has grown exponentially, as did the amount of papers employing deep networks. Vision tasks, such as human pose estimation, did not escape from this trend. There is a large number of deep models, where small changes in the network architecture, or in the data pre-processing, together with the stochastic nature of the optimization procedures, produce notably different results, making extremely difficult to sift methods that significantly outperform others. This situation motivates the current study, in which we perform a systematic evaluation and statistical analysis of vanilla deep regression, i.e., convolutional neural networks with a linear regression top layer. This is the first comprehensive analysis of deep regression techniques. We perform experiments on four vision problems, and report confidence intervals for the median performance as well as the statistical significance of the results, if any. Surprisingly, the variability due to different data pre-processing procedures generally eclipses the variability due to modifications in the network architecture. Our results reinforce the hypothesis according to which, in general, a general-purpose network (e.g., VGG-16 or ResNet-50) adequately tuned can yield results close to the state-of-the-art without having to resort to more complex and ad-hoc regression models.