Search papers, labs, and topics across Lattice.
This paper addresses the underexplored problem of Few-Shot Open-Set Action Recognition (FSOS-AR) in videos by extending existing skeletal-data methods. They introduce a Feature-Residual Discriminator (FR-Disc) module to enhance unknown action rejection without sacrificing closed-set accuracy. Experiments across five datasets demonstrate that FR-Disc outperforms common open-set techniques, establishing a new state-of-the-art for FSOS-AR.
A simple Feature-Residual Discriminator significantly boosts open-set action recognition in videos, where existing techniques fall flat.
Few-Shot Action Recognition (FS-AR) has shown promising results but is often limited by a closed-set assumption that fails in real-world open-set scenarios. While Few-Shot Open-Set (FSOS) recognition is well-established for images, its extension to spatio-temporal video data remains underexplored. To address this, we propose an architectural extension based on a Feature-Residual Discriminator (FR-Disc), adapting previous work on skeletal data to the more complex video domain. Extensive experiments on five datasets demonstrate that while common open-set techniques provide only marginal gains, our FR-Disc significantly enhances unknown rejection capabilities without compromising closed-set accuracy, setting a new state-of-the-art for FSOS-AR. The project website, code, and benchmark are available at: https://hsp-iit.github.io/fsosar/.
