【作者】 谢程利； 王金桥； 卢汉清；

【机构】 中国科学院自动化研究所模式识别国家重点实验室；

【摘要】 计算机视觉往往与机器学习方法一起使用,以处理大量高维数据。随机森林便是一种途径。本文分析了随机森林在视觉任务中的不足,具体而言虽然把样本分散到不同的树中,但并没有对样本维度进行处理;同时由于样本减少而维度未减,给分类工作带来难度。基于此,提出了基于核的随机森林的学习方法,解决了随机森林中(尤其是树的靠后节点)较少样本的分类问题。并且提出将之用于目标检测的实验框架。在VOC数据库上测试,取得了较好的效果。更多还原

【Abstract】 Computer vision usually comes with machine learning, to deal with high dimensional data. Random forest is one kind of such approach. In this paper, the drawback of random forest is pointed out. Specifically, training examples are sampled by different trees. However, the dimension is not reduced. Meanwhile, fewer samples with same dimension make classification more difficult. Thus, a kernel forest method is introduced, to solve the classification problem with fewer samples, especially in the bottom nodes of trees in forest. And then, the object detection application of kernel forest is proposed. Our test on VOC dataset shows good learning model.更多还原