L-LBVC: Long-Term Motion Estimation and Prediction for Learned Bi-Directional Video Compression
This work addresses a critical bottleneck in learned video compression for applications requiring efficient bi-directional coding, representing a strong incremental advance with specific gains.
The paper tackled the problem of inaccurate long-term motion estimation and prediction in learned bi-directional video compression, which lags behind traditional methods, by proposing L-LBVC with adaptive modules for handling motions and reducing bit costs, resulting in significant performance improvements that surpass previous state-of-the-art methods and even VVC on some datasets.
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.