Offline-to-Online Knowledge Distillation for Video Instance Segmentation
This addresses the problem of robust video instance segmentation for challenging datasets with long, dynamic sequences, representing an incremental improvement by combining offline and online models.
The paper tackles video instance segmentation by proposing offline-to-online knowledge distillation to transfer knowledge from an offline model to an online model for consistent prediction, achieving state-of-the-art mAP scores of 46.1%, 43.6%, and 31.1% on YTVIS-21, YTVIS-22, and OVIS datasets.
In this paper, we present offline-to-online knowledge distillation (OOKD) for video instance segmentation (VIS), which transfers a wealth of video knowledge from an offline model to an online model for consistent prediction. Unlike previous methods that having adopting either an online or offline model, our single online model takes advantage of both models by distilling offline knowledge. To transfer knowledge correctly, we propose query filtering and association (QFA), which filters irrelevant queries to exact instances. Our KD with QFA increases the robustness of feature matching by encoding object-centric features from a single frame supplemented by long-range global information. We also propose a simple data augmentation scheme for knowledge distillation in the VIS task that fairly transfers the knowledge of all classes into the online model. Extensive experiments show that our method significantly improves the performance in video instance segmentation, especially for challenging datasets including long, dynamic sequences. Our method also achieves state-of-the-art performance on YTVIS-21, YTVIS-22, and OVIS datasets, with mAP scores of 46.1%, 43.6%, and 31.1%, respectively.