Attributes are an intermediate representation whose purpose is to enable parameter sharing between classes, a must when training data is scarce. We propose to view attribute-based image classification as a label-embedding problem: each class is embedded in the space of attribute vectors. We introduce a function which measures the compatibility between an image and a label embedding. The parameters of this function are learned on a training set of labeled samples to ensure that, given an image, the correct class has a higher compatibility than the incorrect ones. Experimental results on two standard image classification datasets are presented, resp. on the Animals With Attributes and on Caltech-UCSD-Birds datasets.