MambaPSA: A Mamba-based Replacement for C2PSA in YOLO26
Abstract
State space models (SSMs), notably Mamba, have recently emerged as efficient alternatives to self-attention with linear computational complexity. We investigate the integration of Mamba into YOLO26, the latest non-maximum suppression (NMS)-free object detection framework, by proposing MambaPSA, a lightweight Mamba-based replacement for the C2PSA block at the end of the backbone. To complement this study, we additionally insert a bidirectional Vision Mamba (BiViM) module at the P3, P4, and P5 levels of the neck. Experiments on PASCAL VOC 2007+2012 show that MambaPSA reduces parameters by 2.9%, FLOPs by 12.1%, and improves CPU inference throughput by 17.6% (from 17 to 20 FPS) with negligible accuracy change (-0.1 mAP50:95), while the P4 BiViM placement yields the best accuracy gain (+0.9 mAP50:95). These results suggest that SSMs offer a favorable efficiency-accuracy trade-off when replacing attention-based blocks in NMS-free lightweight detectors.
Cite
@article{arxiv.2607.12681,
title = {MambaPSA: A Mamba-based Replacement for C2PSA in YOLO26},
author = {Sheng-Wei Chan and Chia-Min Lin and Hsin-Jui Pan and Ching-Yu Tsai and Chih-Hsiang Yang and Yung-Che Wang and Jen-Shiun Chiang},
journal= {arXiv preprint arXiv:2607.12681},
year = {2026}
}