论文标题
在极端条件下的面部检测：一种机器学习方法
Face Detection in Extreme Conditions: A Machine-learning Approach
论文作者
论文摘要
多年来，由于各种表达，亮度和着色条纹，在不受限制条件下的面部检测一直是一个麻烦。最近的研究表明，对策略的深度学习知识可以在识别不同小工具和模式的识别中获得壮观的表现。由于各种姿势，照明和遮挡，难以限制的环境中的这种面部检测很困难。通过大众媒体普及了弄清楚拥有图片的人。但是，对于指纹或视网膜扫描而言，它不太坚固。最新的研究表明，深层的掌握技术可以在这两个职责上获得令人震惊的表现。在本文中，我建议一个深层级联的多潜水框架，该框架利用它们之间的固有相关性来提高其性能。特别是，我的框架采用了级联的形状，并具有3层谨慎设计的深卷积网络，这些卷积网络期望面部和地标的区域以粗略的方式进行。此外，在获得该过程的知识中，我提出了一种新的在线强硬样本挖掘方法，可以在没有手动模式的情况下机器人增强性能。
Face detection in unrestricted conditions has been a trouble for years due to various expressions, brightness, and coloration fringing. Recent studies show that deep learning knowledge of strategies can acquire spectacular performance inside the identification of different gadgets and patterns. This face detection in unconstrained surroundings is difficult due to various poses, illuminations, and occlusions. Figuring out someone with a picture has been popularized through the mass media. However, it's miles less sturdy to fingerprint or retina scanning. The latest research shows that deep mastering techniques can gain mind-blowing performance on those two responsibilities. In this paper, I recommend a deep cascaded multi-venture framework that exploits the inherent correlation among them to boost up their performance. In particular, my framework adopts a cascaded shape with 3 layers of cautiously designed deep convolutional networks that expect face and landmark region in a coarse-to-fine way. Besides, within the gaining knowledge of the procedure, I propose a new online tough sample mining method that can enhance the performance robotically without manual pattern choice.
