SynthoGestures: A Novel Framework for Synthetic Dynamic Hand Gesture Generation for Driving Scenarios
archive: archived pipeline: cataloged verified
Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)
Summary
This paper introduces SynthoGestures, a framework designed to generate synthetic dynamic hand gestures for automotive human-machine interfaces. The research addresses the high cost and time required to collect diverse, real-world gesture datasets, particularly in dual-task scenarios like driving. Existing synthetic data tools often produce static gestures with limited variation, leading to poor model generalization. SynthoGestures leverages Unreal Engine to create realistic, customizable dynamic gestures, aiming to augment or replace real data to improve recognition accuracy and reduce overfitting. The framework utilizes 3D character models and animation software to simulate gestures across multiple camera types (RGB, infrared, and depth) and viewpoints (e.g., above the driver, behind the wheel). It incorporates realistic noise modeling, such as depth noise based on distance and infrared effects using Fresnel shaders. The system supports two generation modes: description-based, which uses splines to define movement paths, and animation-based. Users can customize parameters including gesture speed, finger spacing, hand orientation, and lighting. The framework automates the iteration through these settings to produce comprehensive datasets with varied performance characteristics. To evaluate the framework, the authors generated 600 synthetic gesture videos (100 variants each) for six specific gestures, using variations in speed, position, and finger rotation. They employed the NVIDIA Dynamic Hand Gesture Dataset and a state-of-the-art recognition model. Experiments compared baseline models trained on real data against those pre-trained on synthetic data or trained on mixed synthetic and real data. Results showed that models trained simultaneously with synthetic and real data achieved the highest accuracy, reaching 89.58%, compared to 79.86% for sequential training and 27.78% for synthetic-only training. The study also analyzed the impact of variation ranges, finding that median ranges for speed, position, and finger spacing yielded optimal performance, while extreme ranges degraded accuracy. The significance of this work lies in providing a cost-effective, flexible tool for generating high-quality synthetic data that enhances gesture recognition models. By demonstrating that synthetic data can significantly boost accuracy when combined with real data, the framework offers a viable solution to dataset biases and scarcity. This approach facilitates faster development of automotive gesture recognition systems by reducing the need for extensive real-world data collection while ensuring robust model generalization through controlled, diverse synthetic variations.
Key finding
Augmenting real hand gesture datasets with synthetic data generated by the SynthoGestures framework significantly improves gesture recognition accuracy, with simultaneous training on both data types yielding the highest performance gains.
Methodology
simulation_modeling
Provenance
The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed. Discovered via openalex_abstract on 2026-05-08 (4 acquisition events logged).
| Stage | Outcome | Tool | Model | Prompt | Attempts | Completed |
|---|---|---|---|---|---|---|
| discover | success | — | — | — | 1 | 2026-05-07 |
| archive | success | unpaywall | — | — | 5 | 2026-06-06 |
| extract | success | cached | — | — | 3 | 2026-06-10 |
| clean | success | clean | — | — | 1 | 2026-06-04 |
| chunk | success | chunk | — | — | 1 | 2026-06-04 |
| embed | success | embed | Qwen/Qwen3-Embedding-8B | — | 1 | 2026-06-04 |
| enrich | success | openalex | — | — | 3 | 2026-05-08 |
| promote | success | — | — | — | 1 | 2026-05-07 |
| summarize | success | llm | qwen3.6-27b-prismaquant | summ-v5 | 2 | 2026-06-10 |
| tag | success | vector_similarity | — | — | 15 | 2026-06-11 |
| verify | success | — | — | — | 2 | 2026-06-10 |
Summary generated by qwen3.6-27b-prismaquant on 2026-06-10; verification: verified.
Topics
Ranked by relevance to this paper. Hover a topic for its definition.
Information type
What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).
- Methodological Resource: tool software