icon

MoRIC: A Modular Region-based Implicit Codec
for Image Compression

Gen Li*, Haotian Wu*, and Deniz Gündüz

*Equal contribution

Imperial College London

NeurIPS 2025

Abstract

We introduce Modular Region-Based Implicit Codec (MoRIC), a novel image compression algorithm that relies on implicit neural representations (INRs). Unlike previous INR-based codecs that model the entire image with a single neural network, MoRIC assigns dedicated models to distinct regions in the image, each tai- lored to its local distribution. This region-wise design enhances adaptation to local statistics and enables flexible, single-object compression with fine-grained rate- distortion (RD) control. MoRIC allows regions of arbitrary shapes, and provides the contour information for each region as separate information. In particular, it in- corporates adaptive chain coding for lossy and lossless contour compression, and a shared global modulator that injects multi-scale global context into local overfitting processes in a coarse-to-fine manner. MoRIC achieves state-of-the-art performance in single-object compression with significantly lower decoding complexity than existing learned neural codecs, which results in a highly efficient compression approach for fixed-background scenarios, e.g., for surveillance cameras. It also sets a new benchmark among overfitted codecs for standard image compression. Additionally, MoRIC naturally supports semantically meaningful layered compres- sion through selective region refinement, paving the way for scalable and flexible INR-based codecs.

Middle illustration

Fig. 1: llustration of the MoRIC framework.

Complexity analysis of the MoRIC

Description of the image

Fig. 2: RD performance for standard image compression. BD-rate vs. decoding complexity on (a) Davis. (b) Kodak, and (c) CLIC2020 datasets.

Workflow of MoRIC Scheme

Description of the image

Fig. 3: The image is partitioned into \( N \) regions, each compressed using a distinct LSN \( g_{\hat{\boldsymbol{W}}^i} \) modulated with a shared GMN \( f_{\hat{\boldsymbol{\theta}}} \).

Illustration of the PCM mechanism

Description of the image

Fig. 4: Red blocks with shared icons indicate parameter-shared components across regions. Each LSN is modulated using multi-scale features generated by \( f_{\hat{\boldsymbol{\theta}}} \) from the corresponding latent vector \( \hat{\boldsymbol{z}}^i \).

Numerical results of LotteryCodec.

Description of the image

Fig. 5: RD performance for standard image compression. (a) RD curve and BD rate on Davis. (b) RD curve and BD rate on Kodak. (c) RD curve and BD rate on CLIC2020.

MoRIC 0.185bpp 35.40dB
C3 0.191bpp 35.10dB
background
local1
local2
No mask selected

BibTeX


  @article{MoRIC,
    title={MoRIC: A Modular Region-based Implicit Codec for Image Compression},
    author={Gen Li, Haotian Wu, and Deniz Gündüz},
    journal={Conference on Neural Information Processing Systems},
    year={2025}
  }