Construct Informative Triplet with Two-stage Hard-sample Generation
This work addresses the challenge of generating effective training samples for metric learning, which is crucial for applications like image retrieval and face recognition, though it appears incremental as it builds on existing hard-sample generation and triplet mining approaches.
The paper tackles the problem of constructing informative triplets for deep metric learning by proposing a two-stage hard-sample generation framework that stretches anchor-positive pairs and uses adaptive constraints to produce hard samples. The method achieves superior performance on benchmark datasets compared to existing algorithms and further boosts performance when combined with existing triplet mining strategies.
In this paper, we propose a robust sample generation scheme to construct informative triplets. The proposed hard sample generation is a two-stage synthesis framework that produces hard samples through effective positive and negative sample generators in two stages, respectively. The first stage stretches the anchor-positive pairs with piecewise linear manipulation and enhances the quality of generated samples by skillfully designing a conditional generative adversarial network to lower the risk of mode collapse. The second stage utilizes an adaptive reverse metric constraint to generate the final hard samples. Extensive experiments on several benchmark datasets verify that our method achieves superior performance than the existing hard-sample generation algorithms. Besides, we also find that our proposed hard sample generation method combining the existing triplet mining strategies can further boost the deep metric learning performance.