Self-Regulated Learning for Egocentric Video Activity Anticipation
This work addresses a key challenge in egocentric vision for applications like assistive technologies, though it is incremental as it builds on existing paradigms with novel regulation mechanisms.
The paper tackles the problem of error accumulation in recursive sequence prediction for egocentric video activity anticipation by proposing a Self-Regulated Learning framework that regulates intermediate representations to emphasize novel information and correlate with past frames, achieving sharp outperformance over state-of-the-art methods on multiple datasets.
Future activity anticipation is a challenging problem in egocentric vision. As a standard future activity anticipation paradigm, recursive sequence prediction suffers from the accumulation of errors. To address this problem, we propose a simple and effective Self-Regulated Learning framework, which aims to regulate the intermediate representation consecutively to produce representation that (a) emphasizes the novel information in the frame of the current time-stamp in contrast to previously observed content, and (b) reflects its correlation with previously observed frames. The former is achieved by minimizing a contrastive loss, and the latter can be achieved by a dynamic reweighing mechanism to attend to informative frames in the observed content with a similarity comparison between feature of the current frame and observed frames. The learned final video representation can be further enhanced by multi-task learning which performs joint feature learning on the target activity labels and the automatically detected action and object class tokens. SRL sharply outperforms existing state-of-the-art in most cases on two egocentric video datasets and two third-person video datasets. Its effectiveness is also verified by the experimental fact that the action and object concepts that support the activity semantics can be accurately identified.