Mutual Modality Trust with Lightweight Reconstruction Regularization for Fine-grained Tire Pattern Recognition

2026-07-27

AI Summary

Existing fine-grained tire tread recognition methods rely on a single visual modality, making it challenging to jointly model spatial and frequency-domain features and leading to overfitting under limited sample sizes. To address these limitations, this work proposes a dual-branch multimodal framework that separately processes surface and indentation images of tires, enabling cross-modal complementary enhancement through a Mutual Modality Trust (M²T) mechanism. Performance is further improved by integrating hierarchical frequency-domain guidance, bandpass filtering decomposition, and a lightweight reconstruction regularizer (LR²). The authors introduce MTire299—the first paired multi-source tire dataset comprising 299 classes and 14,795 image pairs—and demonstrate the effectiveness of their approach on this dataset as well as two public benchmarks, achieving significant gains in recognition accuracy and robustness under few-shot settings.

Abstract

Visual tire recognition serves as a core supporting technique for vehicle safety monitoring, autonomous driving perception and automated automotive maintenance. Existing fine-grained tire recognition techniques suffer from three prominent limitations. They tend to depend on only one visual source, lack the capacity to jointly model spatial and frequency cues for minute tread texture extraction, and suffer severe overfitting given limited annotated tire imagery. This paper proposes a lightweight fine-grained tire pattern recognition method incorporating dual-branch independent inference and enhanced feature fusion to boost recognition performance. The framework employs two task-specialized branches dedicated to tire surface and tread indentation, respectively, to extract modality-specific discriminative features. Each branch conducts independent prediction, while cross-branch feature fusion exploits Mutual Modality Trust (M$^2$T) to realize complementary feature enhancement across two modalities. Besides, a frequency-domain hierarchical guidance module is devised, which leverages bandpass filters to decompose feature maps into high- and low-frequency components and enables fine-grained cross-layer feature modulation. Furthermore, a Lightweight Reconstruction Regularization (LR$^2$) is introduced to retain abundant intrinsic information within feature embeddings, substantially improving feature stability and recognition robustness under limited labeled training data. In addition, we establish a surface-indentation multi-source dataset namely MTire299 for fine-grained tire tread recognition, which covers 299 categories with a total of 14795 paired image samples. Extensive experiments conducted on two public tire datasets validate the superiority and efficacy of the proposed algorithm.

Problem

• fine-grained tire recognition
• multi-modality
• overfitting
• limited annotated data
• texture extraction

Innovation

• Mutual Modality Trust
• Lightweight Reconstruction Regularization
• Fine-grained Recognition
• Frequency-domain Feature Modulation
• Dual-branch Architecture