Neural Stereo Video Compression with Hybrid Disparity Compensation
Updated
Updated · arxiv.org · Jun 29
Neural Stereo Video Compression with Hybrid Disparity Compensation
1 articles · Updated · arxiv.org · Jun 29
Summary
Researchers have developed a neural stereo video compression framework using a hybrid disparity compensation strategy.
The new approach combines explicit pixel displacement with implicit cross-attention, improving efficiency and robustness in reducing cross-view redundancy.
Extensive tests show the framework outperforms traditional and neural methods, with strong results in both automotive and general multi-view video scenarios.