Abstract: The existing frequency-based action recognition methods achieve impressive performance in improving efficiency. However, they ignore the ...

To address this prob- lem, we propose a novel frequency enhancement (FE) block for efficient compressed video action recognition, including a temporal ...

FREQUENCY ENHANCEMENT NETWORK FOR EFFICIENT COMPRESSED VIDEO ACTION RECOGNITION ; Session: TP1.PA: Action Recognition Poster ; Track: Applications of Machine ...

In this paper, we propose a novel faster frequency-domain compressed video action recognition framework (Faster-FCoViAR) as shown in Figure 1, which can ...

Jun 3, 2024 · Frequency Enhancement Network for Efficient Compressed Video Action Recognition. ... Frequency-Domain Compressed Video Action Recognition ...

This work proposes a deep neural network capable of learning straight from compressed video, demonstrating comparable recognition performance to the ...

Action recognition using information in the frequency domain also allows for faster performances, as compressed videos would be sufficient instead of regular ...

this approach is efficient, it requires to decode the frequency domain ... networks for video action recognition,” in IEEE International Confer- ence ...

Apr 25, 2024 · Nevertheless, few existing algorithms consider the compression factor in practical applications. In this paper, we propose an enhanced VSR model ...

First, we train a neural compressor to compress videos. Second, we optionally use augmentation network to transform the compressed space for doing augmentations ...