TextMamba: Scene Text Detector with Mamba

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zhao, Qiyan, Yan, Yue, Wang, Da-Han
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909947636219904
author Zhao, Qiyan
Yan, Yue
Wang, Da-Han
author_facet Zhao, Qiyan
Yan, Yue
Wang, Da-Han
contents In scene text detection, Transformer-based methods have addressed the global feature extraction limitations inherent in traditional convolution neural network-based methods. However, most directly rely on native Transformer attention layers as encoders without evaluating their cross-domain limitations and inherent shortcomings: forgetting important information or focusing on irrelevant representations when modeling long-range dependencies for text detection. The recently proposed state space model Mamba has demonstrated better long-range dependencies modeling through a linear complexity selection mechanism. Therefore, we propose a novel scene text detector based on Mamba that integrates the selection mechanism with attention layers, enhancing the encoder's ability to extract relevant information from long sequences. We adopt the Top\_k algorithm to explicitly select key information and reduce the interference of irrelevant information in Mamba modeling. Additionally, we design a dual-scale feed-forward network and an embedding pyramid enhancement module to facilitate high-dimensional hidden state interactions and multi-scale feature fusion. Our method achieves state-of-the-art or competitive performance on various benchmarks, with F-measures of 89.7\%, 89.2\%, and 78.5\% on CTW1500, TotalText, and ICDAR19ArT, respectively. Codes will be available.
format Preprint
id arxiv_https___arxiv_org_abs_2512_06657
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle TextMamba: Scene Text Detector with Mamba
Zhao, Qiyan
Yan, Yue
Wang, Da-Han
Computer Vision and Pattern Recognition
Artificial Intelligence
In scene text detection, Transformer-based methods have addressed the global feature extraction limitations inherent in traditional convolution neural network-based methods. However, most directly rely on native Transformer attention layers as encoders without evaluating their cross-domain limitations and inherent shortcomings: forgetting important information or focusing on irrelevant representations when modeling long-range dependencies for text detection. The recently proposed state space model Mamba has demonstrated better long-range dependencies modeling through a linear complexity selection mechanism. Therefore, we propose a novel scene text detector based on Mamba that integrates the selection mechanism with attention layers, enhancing the encoder's ability to extract relevant information from long sequences. We adopt the Top\_k algorithm to explicitly select key information and reduce the interference of irrelevant information in Mamba modeling. Additionally, we design a dual-scale feed-forward network and an embedding pyramid enhancement module to facilitate high-dimensional hidden state interactions and multi-scale feature fusion. Our method achieves state-of-the-art or competitive performance on various benchmarks, with F-measures of 89.7\%, 89.2\%, and 78.5\% on CTW1500, TotalText, and ICDAR19ArT, respectively. Codes will be available.
title TextMamba: Scene Text Detector with Mamba
topic Computer Vision and Pattern Recognition
Artificial Intelligence
url https://arxiv.org/abs/2512.06657