International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 11 Issue: 12 | Dec 2024
p-ISSN: 2395-0072
www.irjet.net
Novel Multi-Modal CNN - Fuzzy Based Hand Gestures Recognizing System Shrivarshan N K1, Sahana Mahesh2, Anbuchelvan A P3 1,2&3 Student, School of Computer Science Engineering and Information Systems, SCORE
Vellore Institute of Technology, Vellore, Tamil Nadu, India --------------------------------------------------------------------***--------------------------------------------------------------------The FIS is employed to perform edge detection on input Abstract - This paper presents a novel multi-model
images by applying fuzzy logic-based rules, thereby reducing noise and enhancing the most salient features of hand gestures. The pre-processed, edge-detected images are then utilized to train a CNN model, which focuses on learning discriminative features essential for accurate gesture classification. This hybrid approach aims to improve the robustness and accuracy of hand gesture recognition systems, especially in challenging real-world environments.
neural network and fuzzy-based system for hand gesture recognition, leveraging the combined strengths of Convolutional Neural Networks (CNN) and Fuzzy Inference Systems (FIS). The proposed approach aims to enhance gesture detection accuracy by integrating a fuzzy logic-based edge detection mechanism with a CNN model for improved feature extraction and classification. The Fuzzy Inference System is designed to detect edges in hand gesture images by applying fuzzy rules and membership functions, which reduces the complexity and noise in the input data. The edge-detected images are then fed into the CNN for training, resulting in a more efficient recognition process with improved accuracy and reduced computational costs. Experimental results demonstrate that the proposed system outperforms traditional CNN-based approaches in terms of recognition accuracy and robustness across varying lighting conditions and hand shapes. This innovative combination of CNN and fuzzy logic provides a reliable and efficient solution for hand gesture recognition, with potential applications in human-computer interaction, sign language interpretation, and virtual reality.
The remainder of the paper is structured as follows: Section II discusses related work in the field of hand gesture recognition. Section III describes the proposed multi-model CNN-Fuzzy system, detailing the architecture and the fuzzy logic-based edge detection process. Section IV presents the experimental setup and results, highlighting the performance of the proposed system compared to conventional methods. Finally, Section V concludes the paper with a discussion on future research directions.
2. LITERARY SURVEY Hand gesture recognition (HGR) has been an active research area due to its applications in human-computer interaction, sign language recognition, and automation. Various techniques and models have been developed to enhance the accuracy and robustness of HGR systems, involving different sensors, machine learning algorithms, and deep learning models.
Keywords – Convolutional Neural Networks, Edge Detection, Fuzzy Inference Systems, Hand gesture analysis, Human Computer Interaction
1. INTRODUCTION
Sharma et al. (2023) proposed a time-distance parameterbased HGR system using multiple ultra-wideband (UWB) radars to capture hand gestures. Their approach demonstrated the potential for UWB radars in accurately detecting and recognizing hand movements by leveraging time and distance parameters. This method is beneficial for environments where conventional cameras may face challenges due to lighting or obstructions.
Hand gesture recognition is a critical area of research in human-computer interaction (HCI), offering intuitive and non-invasive methods for users to interact with digital environments. Traditional approaches to hand gesture recognition primarily rely on deep learning techniques, such as Convolutional Neural Networks (CNNs), which have demonstrated significant success in extracting complex features from images. However, CNN-based models often face challenges in handling noise, variations in lighting, and complex backgrounds, which can adversely affect recognition accuracy.
Another study by Sharma et al. (2023) explored the application of machine learning techniques in HGR. They conducted an extensive review of different machine learning methods, including support vector machines, decision trees, and deep learning models, highlighting the advantages and limitations of each approach in
To address these limitations, this paper proposes a novel hand gesture recognition system that combines the capabilities of CNNs with a Fuzzy Inference System (FIS).
© 2024, IRJET
|
Impact Factor value: 8.315
|
ISO 9001:2008 Certified Journal
|
Page 620