A Hybrid Neural Network with Smart Skip Connections for High-Precision, Low-Latency EMG-Based Hand Gesture Recognition

2025-03-12

AI Summary

To address the critical need for high-accuracy, low-latency electromyography (EMG) gesture recognition in prosthetics and human–machine interaction, this work proposes ConSGruNet—a lightweight end-to-end hybrid network integrating convolutional neural networks (CNNs) with gated recurrent units (GRUs), featuring intelligent skip connections that enable dynamic, cross-layer information flow adaptation. Evaluated on the full Ninapro DB1 dataset, ConSGruNet achieves 99.7% classification accuracy, 25 ms inference latency per sample, and a compact parameter count of 3.946 MB. Both Cohen’s kappa and Matthews correlation coefficient approach 1.0, confirming exceptional reliability and generalizability. To our knowledge, this is the first study to jointly optimize accuracy, speed, and model efficiency for fine-grained 53-class EMG gesture recognition—establishing a deployable paradigm for real-time bioelectric signal decoding.

Technology Category

• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Computer Vision: Biometrics, Face, Gesture & Pose

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

Electromyography (EMG) is extensively used in key biomedical areas, such as prosthetics, and assistive and interactive technologies. This paper presents a new hybrid neural network named ConSGruNet for precise and efficient hand gesture recognition. The proposed model comprises convolutional neural networks with smart skip connections in conjunction with a Gated Recurrent Unit (GRU). The proposed model is trained on the complete Ninapro DB1 dataset. The proposed model boasts an accuracy of 99.7% in classifying 53 classes in just 25 milliseconds. In addition to being fast, the proposed model is lightweight with just 3,946 KB in size. Moreover, the proposed model has also been evaluated for the reliability parameters, i.e., Cohen's kappa coefficient, Matthew's correlation coefficient, and confidence intervals. The close to ideal results of these parameters validate the models performance on unseen data.

Problem

• Develops a hybrid neural network for high-precision EMG-based hand gesture recognition.
• Achieves 99.7% accuracy in classifying 53 gestures with low latency.
• Ensures model reliability using statistical measures like Cohen's kappa coefficient.

Innovation

• Hybrid neural network with smart skip connections
• Combines CNN and GRU for gesture recognition
• Achieves 99.7% accuracy in 25 milliseconds