Dynamic sampling for deep metric learning.

• A flexible margin is adopted to filter out easy training samples gradually. • A dynamic sampling strategy is proposed for deep metric learning. • The proposed sampling approach is compatible with several popular loss functions. • With dynamic sampling, performance gap between different loss functions is minor. Deep metric learning maps visually similar images onto nearby locations and visually dissimilar images apart from each other in an embedding manifold. The learning process is mainly based on the supplied image negative and positive training pairs. In this paper, a dynamic sampling strategy is proposed to organize the training pairs in an easy-to-hard order to feed into the network. It allows the network to learn general boundaries between categories from the easy training pairs at its early stages and finalize the details of the model mainly relying on the hard training samples in the later. Compared to the existing training sample mining approaches, the hard samples are mined with little harm to the learned general model. This dynamic sampling strategy is formulated as two simple terms that are compatible with various loss functions. Consistent performance boost is observed when it is integrated with several popular loss functions on fashion search and fine-grained image search. [ABSTRACT FROM AUTHOR]