Part of Advances in Neural Information Processing Systems 35 (NeurIPS 2022) Main Conference Track
Philip de Rijk, Lukas Schneider, Marius Cordts, Dariu Gavrila
Knowledge Distillation (KD) is a well-known training paradigm in deep neural networks where knowledge acquired by a large teacher model is transferred to a small student.KD has proven to be an effective technique to significantly improve the student's performance for various tasks including object detection. As such, KD techniques mostly rely on guidance at the intermediate feature level, which is typically implemented by minimizing an $\ell_{p}$-norm distance between teacher and student activations during training. In this paper, we propose a replacement for the pixel-wise independent $\ell_{p}$-norm based on the structural similarity (SSIM).By taking into account additional contrast and structural cues, more information within intermediate feature maps can be preserved. Extensive experiments on MSCOCO demonstrate the effectiveness of our method across different training schemes and architectures. Our method adds only little computational overhead, is straightforward to implement and at the same time it significantly outperforms the standard $\ell_p$-norms.Moreover, more complex state-of-the-art KD methods using attention-based sampling mechanisms are outperformed, including a +3.5 AP gain using a Faster R-CNN R-50 compared to a vanilla model.