Adaptive Transform Coding for Semantic Compression

TL;DR

This paper introduces an adaptive transform coding approach for semantic image feature compression, improving efficiency and flexibility over existing neural methods by leveraging mode-dependent transforms based on Gaussian mixture models.

Contribution

The paper presents a novel adaptive transform coding scheme that dynamically selects transforms and quantizers according to source components, enhancing semantic feature compression.

Findings

Outperforms or matches state-of-the-art neural compression methods.

Uses mode-dependent transforms for heterogeneous feature distributions.

Maintains flexibility and interpretability in semantic compression.

Abstract

Visual data compression is shifting from human-centered reconstruction to machine-oriented representation coding. In this setting, an image is often mapped to a compact semantic embedding, which is then compressed and transmitted for downstream inference. We propose an adaptive transform-coding method for semantic-feature compression motivated by the conditional rate-distortion function of a Gaussian mixture model. The scheme uses mode-dependent transforms and quantizers selected according to the inferred source component, enabling more efficient coding of heterogeneous feature distributions. Evaluations on features from widely used vision backbones and foundation models show that the proposed method outperforms or is competitive with state-of-the-art neural compression methods while preserving flexibility and interpretability.