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 1705.02498cs.CV
收藏

视觉失真下人类和深度学习识别性能的研究和比较

A Study and Comparison of Human and Deep Learning Recognition Performance Under Visual Distortions

Arizona State University(亚利桑那州立大学)

Samuel Dodge, Lina Karam

摘要:

深度神经网络(DNN)在标准分类任务上实现了出色的性能。然而,在诸如模糊和噪声的图像质量失真下,分类精度变差。在这项工作中,我们比较了DNN与人类受试者在失真图像上的性能。我们表明,尽管DNN在高质量图像上的表现优于或与人类相当,但DNN在失真图像上的表现仍然远低于人类。我们还发现,DNN和人类受试者之间的错误几乎没有相关性。这可能表明DNN和人类视觉系统之间的图像内部表示是不同的。这些与人类表现的比较可以用来指导未来开发更强大的DNN。

英文摘要:

Deep neural networks (DNNs) achieve excellent performance on standard classification tasks. However, under image quality distortions such as blur and noise, classification accuracy becomes poor. In this work, we compare the performance of DNNs with human subjects on distorted images. We show that, although DNNs perform better than or on par with humans on good quality images, DNN performance is still much lower than human performance on distorted images. We additionally find that there is little correlation in errors between DNNs and human subjects. This could be an indication that the internal representation of images are different between DNNs and the human visual system. These comparisons with human performance could be used to guide future development of more robust DNNs.

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。

↑