MFil-Mamba: Multi-Filter Scanning for Spatial Redundancy-Aware Visual State Space Models
Puskal Khadka, KC Santosh Senior, IEEE · 2026
Read MFil-Mamba: Multi-Filter Scanning for Spatial Redundancy-Aware Visual State Space Models by Puskal Khadka, KC Santosh Senior, IEEE (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-20074.txt.
Description
Khadka and Santosh adapt Mamba-style state space models for image tasks with multi-filter scanning. Vision data is not a simple sequence, so plain scans add redundancy and warp spatial links. Their MFil-Mamba backbone gives each scan its own filter. Each pass then gathers distinct, context-rich spatial information. An adaptive weighting step fuses the scan outputs. The model beat leading rivals across several benchmarks. These covered image classification, object detection, and both instance and semantic segmentation. The tiny version also held up well in their tests.
Catalog record
| CREATOR | Puskal Khadka, KC Santosh Senior, IEEE |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 52931 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.20074