L-LBVC: Long-Term Motion Estimation and Prediction for Learned Bi-Directional Video Compression
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866912307459653632 |
|---|---|
| author | Zhai, Yongqi Tang, Luyang Jiang, Wei Yang, Jiayu Wang, Ronggang |
| author_facet | Zhai, Yongqi Tang, Luyang Jiang, Wei Yang, Jiayu Wang, Ronggang |
| contents | Recently, learned video compression (LVC) has shown superior performance under low-delay configuration. However, the performance of learned bi-directional video compression (LBVC) still lags behind traditional bi-directional coding. The performance gap mainly arises from inaccurate long-term motion estimation and prediction of distant frames, especially in large motion scenes. To solve these two critical problems, this paper proposes a novel LBVC framework, namely L-LBVC. Firstly, we propose an adaptive motion estimation module that can handle both short-term and long-term motions. Specifically, we directly estimate the optical flows for adjacent frames and non-adjacent frames with small motions. For non-adjacent frames with large motions, we recursively accumulate local flows between adjacent frames to estimate long-term flows. Secondly, we propose an adaptive motion prediction module that can largely reduce the bit cost for motion coding. To improve the accuracy of long-term motion prediction, we adaptively downsample reference frames during testing to match the motion ranges observed during training. Experiments show that our L-LBVC significantly outperforms previous state-of-the-art LVC methods and even surpasses VVC (VTM) on some test datasets under random access configuration. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2504_02560 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | L-LBVC: Long-Term Motion Estimation and Prediction for Learned Bi-Directional Video Compression Zhai, Yongqi Tang, Luyang Jiang, Wei Yang, Jiayu Wang, Ronggang Computer Vision and Pattern Recognition Multimedia Recently, learned video compression (LVC) has shown superior performance under low-delay configuration. However, the performance of learned bi-directional video compression (LBVC) still lags behind traditional bi-directional coding. The performance gap mainly arises from inaccurate long-term motion estimation and prediction of distant frames, especially in large motion scenes. To solve these two critical problems, this paper proposes a novel LBVC framework, namely L-LBVC. Firstly, we propose an adaptive motion estimation module that can handle both short-term and long-term motions. Specifically, we directly estimate the optical flows for adjacent frames and non-adjacent frames with small motions. For non-adjacent frames with large motions, we recursively accumulate local flows between adjacent frames to estimate long-term flows. Secondly, we propose an adaptive motion prediction module that can largely reduce the bit cost for motion coding. To improve the accuracy of long-term motion prediction, we adaptively downsample reference frames during testing to match the motion ranges observed during training. Experiments show that our L-LBVC significantly outperforms previous state-of-the-art LVC methods and even surpasses VVC (VTM) on some test datasets under random access configuration. |
| title | L-LBVC: Long-Term Motion Estimation and Prediction for Learned Bi-Directional Video Compression |
| topic | Computer Vision and Pattern Recognition Multimedia |
| url | https://arxiv.org/abs/2504.02560 |