SPUDNIG

SPUDNIG automates detection and temporal annotation of iconic and non-iconic hand gestures in video to support quantitative studies of multimodal human communication.

Key Features:

• Automated detection: Detects iconic and non-iconic hand gestures in video and identifies gesture initiation and termination points for precise temporal annotations.
• Annotation compatibility: Produces outputs importable into ELAN and ANVIL for integration with existing annotation workflows.
• Validation: Validated against manual annotations by human coders and shown to reduce the extent of manual annotation while still allowing human oversight for false positives.
• Scalability and consistency: Facilitates large-scale analysis of video corpora and enhances consistency of temporal gesture annotations across datasets.

Scientific Applications:

• Multimodal communication research: Enables analysis of interactions between verbal and non-verbal cues in linguistics, psychology, and communication studies.
• Corpus annotation: Supports large-scale annotation of video corpora for quantitative gesture and interaction analyses.

Methodology:

Analyzes video files to extract hand movements and identifies gesture initiation and termination points for automated temporal annotation.