Towards Flops-Constrained Face Recognition

Large scale face recognition is challenging especially when the computational budget is limited. Given a flops upper bound, the key is to find the optimal neural network architecture and optimization method. In this article, we introduce the solutions of team 'trojans' for the ICCV19 - Lightweight Face Recognition Challenge. Our team mainly focuses on the two 'large' tracks, image-based and video-based, respectively. The submissions of these two tracks are required to be one single model with computational budget no higher than 30 GFlops. We introduce a network architecture 'Efficient PolyFace', a novel loss function 'ArcNegFace', a novel frame aggregation method 'QAN++', together with a bag of useful tricks in our implementation (augmentations, regular face, label smoothing, anchor finetuning, etc.). Our basic model, 'Efficient PolyFace', takes 28.25 Gflops for the 'deepglint-large' image-based track, and the 'PolyFace+QAN++' solution takes 24.12 Gflops for the 'iQiyi-large' video-based track. These two solutions achieve 94.198% @ 1e-8 and 72.981% @ 1e-4 in the two tracks respectively, which are the state-of-the-art results.