An Exploration into Human-Computer Interaction: Hand Gesture Recognition Management in a Challenging Environment.

Area of Science:

• Computer Science
• Human-Computer Interaction
• Artificial Intelligence

Background:

• Speech-impaired individuals often rely on hand gestures for communication, yet are underrepresented in HCI research.
• Existing human-computer interaction systems lack accessibility for the speech-impaired community.
• Developing intuitive interaction methods is crucial for inclusivity.

Purpose of the Study:

• To create an accessible hand gesture recognition system for the speech-impaired community.
• To improve human-computer interaction efficiency and effortlessness without external devices.
• To address the underrepresentation of the speech-impaired in HCI and automation research.

Main Methods:

• A two-phase algorithm involving Region of Interest (ROI) segmentation and Convolutional Neural Network (CNN) image categorization.
• Color space segmentation technique to isolate hand gestures from the background.
• Utilizing Python Keras package for CNN model training and image classification.

Main Results:

• The developed system demonstrated the necessity of image segmentation for effective hand gesture recognition.
• The optimal CNN model achieved a 58% performance accuracy.
• Performance increased by approximately 10% with image segmentation compared to without.

Conclusions:

• Image segmentation is a critical component for improving hand gesture recognition accuracy.
• The developed system offers a pathway towards more inclusive human-computer interaction for the speech-impaired.
• Further research can build upon these findings to enhance accessibility in digital environments.