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

Yun Zhang; Junle Liu; Huan Zhang; Zhaoqing Pan; Gangyi Jiang; Weisi Lin

eess.IVApr 7, 2026

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

Authors: Yun Zhang, Junle Liu, Huan Zhang, Zhaoqing Pan, Gangyi Jiang, Weisi Lin

Organizations: School of Electronics and Communication Engineering, Shenzhen Campus, Sun Yat-Sen University, Shenzhen 518107, China. · School of Information Engineering, Guangdong University of Technology, Guangzhou 510006, China. · School of Electrical and Information Engineering, Tianjin University, Tianjin 300072, China. · Faculty of Electrical Engineering and Computer Science, Ningbo University, Ningbo 315211, China. · College of Computing and Data Science, Nanyang Technological University, Singapore

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.

Figures & tables

Panel of Figure 1 — Fig. 1: Feature channel importance analysis by adding distortions, where mAP@50:95 was used as precision metric for object detection. (a) zero out for individual channel, (b) zero out for a group of channels, (c) distortions added to individual channel, where “Intensity 1” and “Intensity 2” indicated the magnitudes of the added random noise ranges $[-0.5,0.5]$ and $[-1,1]$ , respectively, (d) distortions added to a group of channels.

Panel of Figure 2 — Fig. 2: Framework of the proposed CI-ICM. (a) Network architecture of the proposed CI-ICM, (b) The reordering, grouping, and scaling feature flow of $\mathbf{y}$ , where darker green $\mathbf{y}$ blocks are features with higher importance for machine vision.

Panel of Figure 3 — Fig. 3: Relationship between the number of removed channels and the normalized object detection task accuracy. (a) Normalized mAP@50:95, (b) Normalized mAP@75.

Fig. 4: Decoding process of the CI-CTX module.

Panel of Figure 5 — Fig. 5: $MSE(\mathbf{\Phi}_{\text{ch-org}}^{i},\mathbf{\Phi}_{\text{ch}}^{i})$ with different $s_{i}$ . Note that $s_{i}$ , $i\in[1,2,3]$ are plotted in $1/s_{i}$ , and $s_{4}$ are plotted in $log_{10}(s_{4})$ for better observation. (a) $s_{1}$ , (b) $s_{2}$ , (c) $s_{3}$ , (d) $s_{4}$ .

Panel of Figure 6 — Fig. 6: Object task precisions using different $s_{i}$ , where quadratic functions are used to fit results. Note that $s_{i}$ , $i\in[1,2,3]$ are plotted in $1/s_{i}$ , and $s_{4}$ are plotted in $log_{10}(s_{4})$ for better observation. (a) $s_{1}$ , (b) $s_{2}$ , (c) $s_{3}$ , (d) $s_{4}$ .

Fig. 8: Three stages of training for the proposed CI-ICM, where dark green $\mathbf{y}$ blocks are features with higher importance for machine vision, and dark blue $\mathbf{W}_{c}$ blocks are the higher importance weights.

Panel of Figure 9 — Fig. 9: Trained $\mathbf{W}_{c}$ of an image in COCO2017 before and after training using channel order loss. (a) Before training, (b) After training.

Panel of Figure 10 — Fig. 10: Rate-accuracy curves of the proposed CI-ICM and benchmark schemes on object detection task, (a) mAP@50:95, (b) mAP@50, (c) mAP@75.

Panel of Figure 11 — Fig. 11: Visualization of object detection results using different codecs, where bpp values are also presented. (a) and (f): Ground truth, (b) and (g): ELIC, (c) and (h): TransTIC, (d) and (i): AdaptICMH, (e) and (j): the proposed CI-ICM, (a)-(e): image of a person and skis, (f)-(j): image of an oven and a refrigerator.

Coding schemes	BD-mAP@50:95	BD-mAP@50	BD-mAP@75
ELIC [ 36 ]	0	0	0
JPEG [ 42 ]	-6.798	-9.933	-7.729
VTM [ 4 ]	0.171	-0.656	0.511
MLIC [ 38 ]	1.128	1.543	1.418
CF [ 20 ]	3.783	2.914	3.555
SC2 [ 21 ]	8.745	10.348	8.697

TABLE I: BD-mAPs(

\%

) of the proposed CI-ICM and benchmark schemes for object detection task, where ELIC [ 36 ] is the baseline.

Panel of Figure 13 — Fig. 12: Rate-accuracy curves of the proposed CI-ICM and baseline schemes on instance segmentation task, (a) mAP@50:95, (b) mAP@50, (c) mAP@75.

Panel of Figure 14 — Fig. 13: Visualization of instance segmentation results from coded images, where bpp values also presented. (a) and (f): Ground truth, (b) and (g): ELIC, (c) and (h): TransTIC, (d) and (i): AdaptICMH, (e) and (j): the proposed CI-ICM, (a)-(e): image of person and bicycles, (f)-(j): image of a girl under an umbrella.

Coding schemes	BD-mAP@50:95	BD-mAP@50	BD-mAP@75
ELIC [ 36 ]	0	0	0
JPEG [ 42 ]	-5.969	-9.150	-6.786
VTM [ 4 ]	0.542	0.165	0.562
MLIC [ 38 ]	1.233	1.654	1.299
TransTIC [ 33 ]	10.208	13.376	11.535
AdaptICMH [ 34 ]	11.282	15.221	12.591

TABLE II: BD-mAPs(

\%

) of the proposed CI-ICM and benchmark schemes as comparing with the ELIC [ 36 ] for instance segmentation task.

Panel of Figure 16 — Fig. 14: Rate-accuracy curves of ablation studies on object detection task and instance segmentation task, where the task accuracy is measured with mAP@50:95. (a)-(b) analysis on scaling process in FCGS, CI-CTX, CIG, and channel order loss, (c)-(d) analysis on TSCA modes. (a)(c) object detection, and (b)(d) instance segmentation.

Tasks	Coding schemes	BD-mAP@50:95	BD-mAP@50	BD-mAP@75
Object detection	w/o Scaling	14.566	18.834	16.494
	w/o CI-CTX	11.838	15.882	13.723
	w/o CO	12.742	16.824	14.619
	CI-ICM	16.249	20.924	18.491
Instance segmentation	w/o Scaling	12.650	17.464	14.101
	w/o CI-CTX	10.251	13.797	11.545

TABLE III: BD-mAPs(

\%

) of the ablation studies of the CI-ICM for object detection and instance segmentation, where ELIC [ 36 ] is the anchor.

Tasks	Mode	BD-mAP@50:95	BD-mAP@50	BD-mAP@75
Object detection	Unmatched	12.672	16.938	14.516
Object detection	Matched	16.249	20.924	18.491
Instance segmentation	Unmatched	12.778	18.646	13.984
Instance segmentation	Matched	13.718	19.028	15.244

TABLE IV: BD-mAPs(

\%

) of using TSCA with “matched” and “unmatched” mode for object detection and instance segmentation, where ELIC [ 36 ] is used as anchor.

Panel of Figure 19 — Fig. 15: Rate-accuracy curves of generalization studies, where the task accuracy is measured with mAP@50:95. (a) Analysis on COCO 2017 dataset, Faster R-CNN with ResNet-101 backbone for object detection. (b) Analysis on Pascal VOC 2012 dataset, Faster R-CNN with ResNet-50 backbone for object detection. (c) Analysis on COCO 2017 dataset, Mask R-CNN with ResNet-101 backbone for instance segmentation. (d) Analysis on Cityscapes dataset, Mask R-CNN with ResNet-50 backbone for instance segmentation.

Coding schemes	Object detection		Instance segmentation
Coding schemes	ResNet-101	Pascal VOC	ResNet-101	Cityscapes
ELIC [ 36 ]	0	0	0	0
JPEG [ 42 ]	-7.130	-11.267	-6.393	-3.934
VTM [ 4 ]	-2.190	-3.910	-2.076	0.686
MLIC [ 38 ]	1.207	2.112	1.025	2.239
TransTIC [ 33 ]	9.277	15.294	7.491	2.507

TABLE V: BD-mAP@50:95s(

\%

) of the proposed CI-ICM and benchmark schemes for different datasets and feature extraction network, where ELIC [ 36 ] is used as the anchor.

Coding schemes	Coding time (s)		FLOPs (G)	Params. (M)
Coding schemes	Encoding	Decoding	FLOPs (G)	Params. (M)
ELIC [ 36 ]	2.722	0.124	871.420	31.656
MLIC [ 38 ]	2.133	0.203	1317.803	116.481
CF [ 20 ]	2.746	0.251	180.672	34.755
SC2 [ 21 ]	2.401	0.169	151.540	34.014
TransTIC [ 33 ]	2.311	0.175	529.876	9.028

TABLE VI: FLOPs and number of parameters comparison between the proposed CI-ICM and learning-based methods

Explore similar work

You've Seen Enough: Quality-Constrained Image Coding for Machines
Sep 20, 2026Khoa Pham-Dinh, Sanaz Nami, Hamed Rezazadegan Tavakoli +2Learned Image Compression Rate-Distortion Optimization
Tampere University · Nokia Technologies
MoECodec: Image Compression for joint human and machine perception via Mixture-of-Experts
Jun 19, 2026Jiancheng Zhao, Xiang Ji, Yifan Zhan +2Learned Image Compression Mixture-of-Experts Models
The University of Tokyo
What and Where to Adapt: Structure-Semantics Co-Tuning for Machine Vision Compression via Synergistic Adapters
Apr 11, 2026Shaobo Liu, Haobo Xiong, Kai Liu +1Learned Image Compression Adapter Tuning
Xidian University

Sep 20, 2026J/K move · Enter open · S save

You've Seen Enough: Quality-Constrained Image Coding for Machines

Khoa Pham-Dinh, Sanaz Nami, Hamed Rezazadegan Tavakoli, Moncef Gabbouj+1

Tampere University · Nokia Technologies

Learned Image Compression Rate-Distortion Optimization

Open paper AskPDF arXiv

Visual data is increasingly consumed by machine-vision systems rather than by human observers. Image Coding for Machines (ICM) compresses images assuming the main observer is a computer vision application and that the human observer needs to inspect or validate the decisions. Inspired by just-noticeable distortion, we cap human-observed quality at a desired level and devote the remaining bits to machine performance. Specifically, joint compression-segmentation training is recast as a constrained optimization problem in which the codec must meet a predefined acceptable target visual quality while a task term consumes the remaining coding capacity. This paper proposes two variants of a penalty function that guides the quality toward the target: an absolute function and a bilinear function, the latter applying a steeper slope once the target visual quality is exceeded. Experimental results show that, under the quality constraint, the proposed method achieves BD-rates of

-22.82\%

and

-29.81\%

relative to an unconstrained joint rate--distortion--task optimization and a simple rate--distortion baseline, respectively, showcasing bitrate reduction with the same task performance. This is achieved while the codec also meets the target visual quality with a reasonable error and without adding any complexity overhead.

Abstract

Explore similar work

You've Seen Enough: Quality-Constrained Image Coding for Machines

MoECodec: Image Compression for joint human and machine perception via Mixture-of-Experts

What and Where to Adapt: Structure-Semantics Co-Tuning for Machine Vision Compression via Synergistic Adapters