[arxiv 1805] Why do deep convolutional networks generalize so poorly to small image transformations? Aharon Azulay and YairWeiss from Hebrew University of Jerusalem paper link Introduction 深度卷积网络
RoI Tanh-polar Transformer Network for Face Parsing in the Wild 野外人脸解析的RoI(感兴趣区域) Tanh-polar Transformer网络 摘要 人脸解析人脸解析的目的是预测图像中目标人脸的面部成分的像素级标签,现有的方法通常根据预处理(人脸检测模型:MTCNN (TensorFlow)SSD Face (Caff