Published In
Proceedings of the 7th ACM International Conference on Multimedia in Asia Mmasia 2025
Document Type
Conference Proceeding
Publication Date
12-6-2025
Subjects
Video compression, video frame interpolation
Abstract
Traditional video compression continues to advance, but the gainsin efficiency are diminishing and come at the cost of higher compu-tational complexity. Despite achieving competitive rate-distortionresults, current neural video codecs (NVCs) generally lack sup-port for a wide range of quality levels, often requiring multiplemodels to achieve flexible rate control, which increases both train-ing cost and deployment complexity. To address the limitations ofboth traditional codecs and current NVCs, we propose a hybridvideo compression framework that integrates traditional codecswith hint-guided video frame interpolation (VFI), a learning-basedtechnique for synthesizing intermediate frames. By using decodedreference frames and leveraging compressed-domain hints to guideinterpolation, our method improves both motion compensationand reconstruction quality. This design combines the efficiencyof traditional codecs with the adaptability of neural interpolation,achieving consistent rate-distortion performance and supporting awide range of quality levels on standard benchmarks.
Rights
Copyright (c) 2025 The Authors
This work is licensed under a Creative Commons Attribution 4.0 International License.
DOI
10.1145/3743093.3771081
Persistent Identifier
https://archives.pdx.edu/ds/psu/44390
Citation Details
Tan, P., & Feng, W. (2025). Hint-Guided Video Frame Interpolation for Video Compression. Proceedings of the 7th ACM International Conference on Multimedia in Asia, 1–7.