aiqbench analysis
This research introduces CrossMambaTuning, a parameter-efficient fine-tuning framework that combines State Space Models with cross-layer coordination to adapt learned image compression models for machine vision tasks. By integrating Mamba adapters and scale-invariant mechanisms, the method significantly reduces retraining overhead while capturing both local and global feature dependencies across different network layers. The approach achieves state-of-the-art performance on multiple downstream tasks, offering a practical solution for deploying compressed vision models without substantial computational cost.
CrossMambaTuning: Synergistic Spatial and Cross-Layer Adaptation for Machine Vision Compression
arXiv:2608.25568v1 Announce Type: cross Abstract: To reduce deployment cost and retraining overhead, adapting pretrained learned image compression (LIC) models to downstream machine vision tasks has attracted growing attention. However, existing methods typically insert fine-tuning modules independently into frozen backbones, lacking explicit mechanisms for cross-layer coordination. To address this limitation, we propose a novel framework named CrossMambaTuning, which integrates State Space Models with cross-layer interaction mechanisms for parameter-efficient fine-tuning. Specifically, we design an efficient Mamba adapter equipped with task-specific prompts and multi-scale branching to precisely capture both local features and global dependencies. Furthermore, we introduce a Scale-Invariant Cross-Layer Adapter (SICA) utilizing a parameter-sharing strategy to fuse task information across different scales and reduce redundancy. Extensive experiments demonstrate that CrossMambaTuning achieves state-of-the-art (SOTA) performance on multiple machine vision tasks, reducing parameter overhead by 72\% compared to SOTA methods. Code is available at https://github.com/rsr1123/CrossMambaTuning.
Abstract:To reduce deployment cost and retraining overhead, adapting pretrained learned image compression (LIC) models to downstream machine vision tasks has attracted growing attention. However, existing methods typically insert fine-tuning modules independently into frozen backbones, lacking explicit mechanisms for cross-layer coordination. To address this limitation, we propose a novel framework named CrossMambaTuning, which integrates State Space Models with cross-layer interaction mechanisms for parameter-efficient fine-tuning. Specifically, we design an efficient Mamba adapter equipped with task-specific prompts and multi-scale branching to precisely capture both local features and global dependencies. Furthermore, we introduce a Scale-Invariant Cross-Layer Adapter (SICA) utilizing a parameter-sharing strategy to fuse task information across different scales and reduce redundancy. Extensive experiments demonstrate that CrossMambaTuning achieves state-of-the-art (SOTA) performance on multiple machine vision tasks, reducing parameter overhead by 72\% compared to SOTA methods. Code is available at this https URL.
Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
arXiv:2608.25568 [cs.CV]
(or arXiv:2608.25568v1 [cs.CV] for this version)
https://doi.org/10.48550/arXiv.2608.25568 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 236 words
Related coverage
Structured Affinity for Unsupervised Visual Class-Incremental Memory in Deep Artificial Immune Networks
arXiv · Aug 21, 2026
ZOTTA: Test-Time Adaptation with Gradient-Free Zeroth-Order Optimization
arXiv · Aug 25, 2026
Token-Oriented Semantic Communication with Pretrained Vision Transformers
arXiv · Aug 27, 2026
Google DeepMind Ships Three Physical AI Models For Whole Body Control, Dexterity And Multi Robot Collaboration
MarkTechPost · Jul 30, 2026