基于数据均衡的增进式深度自动图像标注

自动图像标注是一个包含众多标签、多样特征的富有挑战性的研究问题，是新一代图像检索与图像理解的关键步骤.针对传统基于浅层机器学习标注算法标注效率低下、难以处理复杂分类任务的问题，本文提出了基于栈式自动编码器（SAE）的自动图像标注算法，提升了标注效率和标注效果.全文主要针对图像标注数据不平衡问题，提出两种解决思路：对于标注模型，我们提出一种增强训练中低频标签的平衡栈式自动编码器（B-SAE），较好地改善了中低频标签的标注效果.并在此模型基础上提出一种分组强化训练B-SAE子模型的鲁棒平衡栈式自动编码器算法（RB-SAE），提升了标注的稳定性，从而保证模型本身具有较强地处理不平衡数据的能力；对于标注过程，我们以未知图像作为出发点，首先构造未知图像的局部均衡数据集，并判定该图像的高低频属性来决定不同的标注过程，局部语义传播算法（SP）标注中低频图像，RB-SAE算法标注高频图像，形成属性判别的标注框架（ADA），保证了标注过程具有较强地应对不平衡数据的能力，从而提升整体图像标注效果.通过在三个公共数据集上进行实验验证，结果表明，本文方法在许多指标上相比以往方法均有较大提高.

Enhanced Deep Automatic Image Annotation Based on Data Equalization

Automatic image annotation is a challenging research problem containing a lot of tags and various features. Aiming at the problem that the image annotation based on the traditional shallow machine learning algorithm has low efficiency and difficult to deal with the complex classification task, this paper proposes an automatic image annotation algorithm based on stacked auto-encoder (SAE) to improve both efficiency and effectiveness of annotation. In this paper, two kinds of strategies are proposed to solve the main problem of unbalanced data in image annotation:For the annotation model itself, to improve the annotation effect of low frequency tags, we propose a balanced and stacked auto-encoder (B-SAE) that can enhance training for low frequency tags. On the basis of this model, a robust balanced and stacked auto-encoder algorithm (RB-SAE) with enhancing training by group for sub B-SAE model is proposed to enhance the annotation stability. This strategy ensures that the model itself has a strong ability to deal with the unbalanced data. For the annotation process, we take the unknown image as the starting point. Then we construct the local equilibrium dataset of the unknown image and discriminate the high-and low-frequency attribute of the image to determine the different annotation process. The local semantic propagation algorithm (SP) annotates the low frequency images and the RB-SAE algorithm annotates the high frequency images. Then the framework of attribute discrimination annotation (ADA) is formed to improve the overall image annotation effect. This strategy ensures that the labeling process has a strong ability to deal with unbalanced data. Experimental results generated from three public data sets show that many indicators in our model compared to the previous models are all improved.