Convert landmark list to Hand21 schema for hand gesture analysis

Use This When

• Standardizing hand landmarks to COCO WholeBody 21-keypoint hand schema
• Building gesture recognition systems requiring finger joint positions
• Interfacing with hand pose analyzers expecting named finger keypoint fields
• Preparing hand landmark data for sign language recognition or hand tracking

What It Does

• Converts ordered 21-point landmark list to structured Hand21 format with named fields
• Maps wrist and 4 joints per finger (thumb, forefinger, middle, ring, pinky)
• Validates minimum landmark count and fails if fewer than 21 landmarks provided
• Returns typed Landmarks2d.Hand21 structure for downstream hand analysis components

Works Best With

• Hand pose detector → this component → gesture recognition or sign language analysis
• detect-keypoints-hf → this component → hand-specific analytics
• Integration with visualize for hand skeleton rendering
• Feeding gesture classifiers that rely on named finger joint access

Caveats

• Requires exactly 21 or more landmarks; fails with fewer even if partial hand acceptable
• No validation that landmark semantics match expected hand ordering; assumes correct sequence
• Confidence scores preserved but not validated; downstream must filter low-confidence joints
• Additional landmarks beyond 21 ignored; uses only first 21 points in order