Abstract Object co-segmentation aims to segment the common objects in images. This paper presents a CNNbased method that is unsupervised and end-to-end trainable to better solve this task. Our method is unsupervised in the sense that it does not require any training data in the form of object masks but merely a set of images jointly covering objects of a specifific class. Our method comprises two collaborative CNN modules, a feature extractor and a co-attention map generator. The former module extracts the features of the estimated objects and backgrounds, and is derived based on the proposed co-attention loss, which minimizes interimage object discrepancy while maximizing intraimage fifigure-ground separation. The latter module is learned to generate co-attention maps by which the estimated fifigure-ground segmentation can better fifit the former module. Besides the co-attention loss, the mask loss is developed to retain the whole objects and remove noises. Experiments show that our method achieves superior results, even outperforming the state-of-the-art, supervised methods.