Kakeru Koizumi

Taiga Hayami, Kakeru Koizumi, Hiroshi Watanabe

Implicit Neural Representations (INRs) have garnered significant attention for their ability to model complex signals in various domains. Recently, INR-based frameworks have shown promise in neural video compression by embedding video content into compact neural networks. However, these methods often struggle to reconstruct high-frequency details under stringent constraints on model size, which are critical in practical compression scenarios. To address this limitation, we propose an INR-based video representation framework that integrates a general-purpose super-resolution (SR) network. This design is motivated by the observation that high-frequency components tend to exhibit low temporal redundancy across frames. By offloading the reconstruction of fine details to a dedicated SR network pre-trained on natural images, the proposed method improves visual fidelity. Experimental results demonstrate that the proposed method outperforms conventional INR-based baselines in reconstruction quality, while maintaining a comparable model size.

Taiga Hayami, Kakeru Koizumi, Hiroshi Watanabe

Implicit neural representations (INRs) are the subject of extensive research, particularly in their application to modeling complex signals by mapping spatial and temporal coordinates to corresponding values. When handling videos, mapping compact inputs to entire frames or spatially partitioned patch images is an effective approach. This strategy better preserves spatial relationships, reduces computational overhead, and improves reconstruction quality compared to coordinate-based mapping. However, predicting entire frames often limits the reconstruction of high-frequency visual details. Additionally, conventional patch-based approaches based on uniform spatial partitioning tend to introduce boundary discontinuities that degrade spatial coherence. We propose a neural video representation method based on Structure-Preserving Patches (SPPs) to address such limitations. Our method separates each video frame into patch images of spatially aligned frames through a deterministic pixel-based splitting similar to PixelUnshuffle. This operation preserves the global spatial structure while allowing patch-level decoding. We train the decoder to reconstruct these structured patches, enabling a global-to-local decoding strategy that captures the global layout first and refines local details. This effectively reduces boundary artifacts and mitigates distortions from naive upsampling. Experiments on standard video datasets demonstrate that our method achieves higher reconstruction quality and better compression performance than existing INR-based baselines.