CI-ICM: Channel Importance-driven Learned Image Coding for Machines

2026-04-06

AI Summary

This work addresses the limitation of traditional image compression methods, which are optimized for human visual perception and often fail to preserve features critical for machine vision tasks. To bridge this gap, the authors propose a novel machine vision-oriented image coding framework that introduces, for the first time, a channel importance-driven mechanism. This mechanism comprises a Channel Importance Generation (CIG) module to quantify the contribution of feature channels, integrated with Non-uniform Grouping and Scaling (FCGS), Channel Importance-aware Context Modeling (CI-CTX), and Task-adaptive Channel Adaptation (TSCA) to jointly optimize rate-distortion performance and downstream task accuracy. Evaluated on COCO2017, the proposed method achieves a 16.25% gain in BD-mAP@50:95 for object detection and a 13.72% improvement for instance segmentation over the baseline, significantly narrowing the divide between image compression and machine perception.

Abstract

Traditional human vision-centric image compression methods are suboptimal for machine vision centric compression due to different visual properties and feature characteristics. To address this problem, we propose a Channel Importance-driven learned Image Coding for Machines (CI-ICM), aiming to maximize the performance of machine vision tasks at a given bitrate constraint. First, we propose a Channel Importance Generation (CIG) module to quantify channel importance in machine vision and develop a channel order loss to rank channels in descending order. Second, to properly allocate bitrate among feature channels, we propose a Feature Channel Grouping and Scaling (FCGS) module that non-uniformly groups the feature channels based on their importance and adjusts the dynamic range of each group. Based on FCGS, we further propose a Channel Importance-based Context (CI-CTX) module to allocate bits among feature groups and to preserve higher fidelity in critical channels. Third, to adapt to multiple machine tasks, we propose a Task-Specific Channel Adaptation (TSCA) module to adaptively enhance features for multiple downstream machine tasks. Experimental results on the COCO2017 dataset show that the proposed CI-ICM achieves BD-mAP@50:95 gains of 16.25$\%$ in object detection and 13.72$\%$ in instance segmentation over the established baseline codec. Ablation studies validate the effectiveness of each contribution, and computation complexity analysis reveals the practicability of the CI-ICM. This work establishes feature channel optimization for machine vision-centric compression, bridging the gap between image coding and machine perception.

Problem

• machine vision
• image compression
• channel importance
• learned image coding
• bitrate allocation

Innovation

• Channel Importance
• Learned Image Coding
• Machine Vision Compression
• Feature Channel Grouping
• Task-Specific Adaptation