Abstract: Hand gesture recognition is a challenging problem in computer vision, with various applications in fields such as robotics, human-computer interaction, and sign language recognition. The ability to recognize hand gestures in real-time can enable seamless communication between humans and machines, making human-computer interaction more intuitive and natural. In this project, we propose a system that can recognize hand gestures by classifying the fingers using open-source computer vision technology. The proposed system uses a combination of image processing techniques and machine learning algorithms to classify the fingers and recognize hand gestures. The input data is captured from a webcam and pre-processed using techniques such as skin colour detection and hand tracking. The hand is then segmented, and the fingers are extracted based on their location and orientation. The extracted finger images are then classified using a convolutional neural network (CNN) architecture. The CNN model is trained on a large dataset of finger images and hand gestures to achieve high accuracy in classification. The dataset comprises images of various hand gestures, including open, closed, and partially closed hands. The CNN model is trained to recognize the fingers' positions and orientations in the input images and classify them into their respective categories. The model is fine-tuned using transfer learning techniques to improve its accuracy and generalizability. The proposed system is evaluated using a variety of hand gestures, including thumbs up, thumbs down, okay, and rock-on, among others. The system achieves high accuracy in recognizing different hand gestures in real-time, with an overall accuracy of 95%. The proposed system's robustness and accuracy make it suitable for various applications, including sign language recognition, human-computer interaction, and gaming. In conclusion, this project proposes a system that can recognize hand gestures by classifying the fingers using open-source computer vision technology. The system achieves high accuracy in real-time hand gesture recognition and has potential applications in various fields. Future work can explore the use of deep learning algorithms and more extensive datasets to improve the system's accuracy and performance. Additionally, the proposed system can be extended to recognize hand gestures in different lighting conditions and backgrounds
|
DOI:
10.17148/IJARCCE.2023.12518
