Answers are grounded in this paper and cite their sources.
Manipulative Hand Gesture Recognition (HGR) focuses on understanding hand-object interactions and assigning semantic labels to each gesture. These techniques play a key role in applications such as robotics, virtual and augmented reality, human–computer interaction, and industrial automation. Hand gestures can be captured using diverse sensor modalities, including RGB, depth, skeleton, infrared, thermal, inertial measurement units (IMUs), surface electromyography (sEMG), and electromagnetic sensors, each with distinct advantages depending on the use case. Based on these data modalities, a wide range of models have been developed to recognize manipulative gestures using single or multiple modalities. In this study, we provide a comprehensive review of recent models in manipulative HGR, focusing on various data modalities. We first analyze the characteristics and taxonomy of manipulative gestures and review commonly used sensors and datasets. Then, we explore deep learning-based recognition methods, including single-modality approaches and multi-modality strategies such as data-level, feature-level, and score-level fusion, as well as co-learning frameworks. Lastly, we present current challenges and promising directions for future research. This study aims to present the first comprehensive review of manipulative HGR, including innovative technologies, methods, and research outcomes, and to analyze the benefits and limitations of various hand gesture recognition algorithms, aiming to contribute to future advancements in the field.
No takes yet. Share an insight, caveat, or question.
Wang et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: