Hand Landmark Data for Sign Language Gestures

Landmark data for hand gestures, formatted for Graph Convolutional Network (GCN) input. The dataset is intended for sign language recognition tasks, specifically focusing on the AlphaSign gesture set. The author, organization, and dataset size are not specified.

Use Cases

• Train a Graph Neural Network to classify AlphaSign gestures using the provided hand landmark graph structure.
• Benchmark GCN architectures for spatial-temporal gesture recognition on sequences of hand landmark graphs.
• Develop a sign language translation model by processing the landmark features as node attributes in a graph.
• Fine-tune pre-trained GCN models for specific hand gesture recognition tasks using this landmark dataset.

Strengths

• Data is pre-processed and structured specifically for Graph Neural Network consumption, saving implementation time.
• Focuses on a defined gesture set (AlphaSign) for sign language recognition.

Limitations

• The total number of gesture samples, subjects, and landmark points per frame is unknown.
• Potential for limited gesture diversity if the AlphaSign set is small or constrained.
• Lacks metadata such as subject demographics or recording conditions.

Provenance

Source

Kaggle

Collection Method

Presumably derived from video or image data of hand gestures, with landmarks extracted and formatted into graphs.