XKD: Cross-modal Knowledge Distillation with Domain Alignment for Video Representation Learning
This work addresses the challenge of multimodal representation learning for video analysis, offering incremental improvements in performance for tasks like action and sound classification.
The paper tackles the problem of learning meaningful representations from unlabelled videos by proposing XKD, a self-supervised framework that uses cross-modal knowledge distillation with domain alignment, resulting in improvements of 8% to 14% in video action classification and achieving 96.5% top-1 accuracy in sound classification.
We present XKD, a novel self-supervised framework to learn meaningful representations from unlabelled videos. XKD is trained with two pseudo objectives. First, masked data reconstruction is performed to learn modality-specific representations from audio and visual streams. Next, self-supervised cross-modal knowledge distillation is performed between the two modalities through a teacher-student setup to learn complementary information. We introduce a novel domain alignment strategy to tackle domain discrepancy between audio and visual modalities enabling effective cross-modal knowledge distillation. Additionally, to develop a general-purpose network capable of handling both audio and visual streams, modality-agnostic variants of XKD are introduced, which use the same pretrained backbone for different audio and visual tasks. Our proposed cross-modal knowledge distillation improves video action classification by $8\%$ to $14\%$ on UCF101, HMDB51, and Kinetics400. Additionally, XKD improves multimodal action classification by $5.5\%$ on Kinetics-Sound. XKD shows state-of-the-art performance in sound classification on ESC50, achieving top-1 accuracy of $96.5\%$.