A Survey on Hand Pose Estimation with Wearable Sensors and Computer-Vision-Based Methods.

Area of Science:

• Computer Science
• Robotics
• Human-Computer Interaction

Background:

• Real-time human hand sensing and modeling are crucial for natural human-computer interactions.
• Hand pose estimation presents significant academic and technical challenges due to hand complexity and dexterity.
• Recent advancements in hardware and AI have spurred development in data gloves and computer-vision methods.

Purpose of the Study:

• To provide a comprehensive and timely review of sensor-based hand pose estimation techniques.
• To cover both wearable (data gloves) and vision-based solutions.
• To discuss the role of deep learning in advancing hand pose estimation.

Main Methods:

• Review of existing literature on sensor-based hand pose estimation.
• Discussion of hand kinematic models.
• In-depth analysis of data gloves and vision-based sensor systems, including their modeling approaches.
• Focus on deep-learning-based methods for hand pose estimation.

Main Results:

• Existing reviews are often limited in scope (e.g., focusing only on data gloves or specific camera types).
• Sensor-based methods, including data gloves and vision systems, offer diverse approaches to hand pose estimation.
• Deep learning methods show significant promise for improving accuracy and efficiency.

Conclusions:

• A comprehensive review of sensor-based hand pose estimation is needed.
• Both wearable and vision-based sensors, especially those leveraging deep learning, are key areas of advancement.
• Further discussion on advantages, drawbacks, applications, and challenges is warranted.