MoExDA: Domain Adaptation for Edge-based Action Recognition
This work addresses static bias for action recognition systems, offering an incremental improvement in generalization.
The paper tackles the problem of static bias in action recognition models by proposing MoExDA, a lightweight domain adaptation method that uses edge frames alongside RGB frames, resulting in more robust action recognition with lower computational cost.
Modern action recognition models suffer from static bias, leading to reduced generalization performance. In this paper, we propose MoExDA, a lightweight domain adaptation between RGB and edge information using edge frames in addition to RGB frames to counter the static bias issue. Experiments demonstrate that the proposed method effectively suppresses static bias with a lower computational cost, allowing for more robust action recognition than previous approaches.