G-MSGINet: A Grouped Multi-Scale Graph-Involution Network for Contactless Fingerprint Recognition

TL;DR

G-MSGINet is presented, a unified and efficient framework for robust contactless fingerprint recognition that jointly performs minutiae localization and identity embedding directly from raw input images and eliminates explicit orientation supervision and adapts graph connectivity directly from learned kernel descriptors, thereby capturing meaningful structural relationships among fingerprint regions without fixed heuristics.

Abstract

This paper presents G-MSGINet, a unified and efficient framework for robust contactless fingerprint recognition that jointly performs minutiae localization and identity embedding directly from raw input images. Existing approaches rely on multi-branch architectures, orientation labels, or complex preprocessing steps, which limit scalability and generalization across real-world acquisition scenarios. In contrast, the proposed architecture introduces the GMSGI layer, a novel computational module that integrates grouped pixel-level involution, dynamic multi-scale kernel generation, and graph-based relational modelling into a single processing unit. Stacked GMSGI layers progressively refine both local minutiae-sensitive features and global topological representations through end-to-end optimization. The architecture eliminates explicit orientation supervision and adapts graph connectivity directly from learned kernel descriptors, thereby capturing meaningful structural relationships among fingerprint regions without fixed heuristics. Extensive experiments on three benchmark datasets, namely PolyU, CFPose, and Benchmark 2D/3D, demonstrate that G-MSGINet consistently achieves minutiae F1-scores in the range of $0.83\pm0.02$ and Rank-1 identification accuracies between 97.0% and 99.1%, while maintaining an Equal Error Rate (EER) as low as 0.5%. These results correspond to improvements of up to 4.8% in F1-score and 1.4% in Rank-1 accuracy when compared to prior methods, using only 0.38 million parameters and 6.63 giga floating-point operations, which represents up to ten times fewer parameters than competitive baselines. This highlights the scalability and effectiveness of G-MSGINet in real-world contactless biometric recognition scenarios.

Figures (6)

• Figure 1: Overview of the proposed method. The network processes a full fingerprint image to generate a minutiae heatmap and an identity embedding. Numbered blocks correspond to key components, with section numbers indicating where each is described.
• Figure 2: Illustration of the G-MSGI layer. The input tensor of shape $C \times H \times W$ is divided into spatial groups (e.g., $2 \times 2$, shown in different colors). Each group is processed by $n$ multi-scale kernel generators $\{\phi_i\}_{i=1}^n$, producing kernels that are applied across all pixels in the group (visualized here for a single pixel in pink for simplicity), and their outputs are averaged to obtain the involution result. Simultaneously, node features for each group are computed via $\phi_{\text{graph}}$, and each kernel is projected using a corresponding projection function $\{\phi_{\text{proj}_i}\}$. These projected kernels are averaged to form node descriptors (colored to indicate node identity), which are used to compute the adjacency matrix $A$ for graph construction. Graph message passing is then performed over these nodes to obtain a graph-enhanced feature representation. Circles marked with $\otimes$ and $\oplus$ denote element-wise multiplication and addition, respectively.
• Figure 3: G-MSGINet architecture. The input image is processed through convolutional and MSGI layers (in yellow), producing two outputs: a $400 \times 400$ minutiae heatmap and a 128-dimensional feature vector.
• Figure 4: Predicted minutiae points (red circles) overlaid on sample images from CFPose, Benchmark 2D/3D, and PolyU datasets (top to bottom, left to right).
• Figure 5: Evaluation across three datasets: Cosine similarity distributions highlight intra- vs. inter-class feature separability, while F1-score comparisons demonstrate the superior discriminative performance of G-MSGINet.