Go to ICLR 2025 Conference homepageFV-NeRV: Neural Compression for Free Viewpoint Videos
Keywords: Implicit Neural Representation, Free Viewpoint Video
Abstract: The delivery of free viewpoint videos (FVVs) is gaining popularity because of their ability to provide freely switchable perspectives to remote users as immersive experiences. While smooth view switching is crucial for enhancing user's experiences, FVV delivery faces a significant challenge in balancing traffic and decoding latency. The typical approach sends limited viewpoints and synthesizes the remainings on the user, reducing traffic, but increasing decoding delays. Alternatively, sending more viewpoints reduces the delay, but requires more bandwidth for transmission. In this paper, we propose a novel FVV representation format, Free Viewpoint-Neural Representation for Videos~(FV-NeRV), to address this dilemma in FVV delivery. FV-NeRV reduces both traffic and decoding delay even for content with a large number of virtual viewpoints by overfitting compact neural networks to all viewpoints and pruning and quantizing the trained model. Experiments show that FV-NeRV achieves a comparable or even superior traffic reduction with a faster decoding speed compared to existing FVV codecs.
Primary Area: applications to computer vision, audio, language, and other modalities
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.
Reciprocal Reviewing: I understand the reciprocal reviewing requirement as described on https://iclr.cc/Conferences/2025/CallForPapers. If none of the authors are registered as a reviewer, it may result in a desk rejection at the discretion of the program chairs. To request an exception, please complete this form at https://forms.gle/Huojr6VjkFxiQsUp6.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 4567
Loading