AI Engineer Needed for Driver Behavior Classification Using 3D Skeleton Data

Job Title:
AI Engineer: ST-GCN Driver Behavior Classifier + ProtoGCN Prototypes (3D Skeleton Data)
Description:
Seeking an experienced AI/Deep Learning Engineer to build a two-phase framework for driver behavior analysis using 3D skeleton data.
Input Data: CSV files with time-indexed 3D joint coordinates (XYZ positions, 15-25 joints, frames/sequence). Sample provided. About 2,000 dataset across behaviors: normal driving, drinking, phone use, eye rubbing, etc.
Phase 1: ST-GCN Classifier
- Preprocess skeleton sequences (normalization, padding, graph construction).
- Implement ST-GCN for multi-class driver behavior classification.
- Train/evaluate on provided dataset; report accuracy, F1, confusion matrix.
Phase 2: ProtoGCN Prototypes
- Extract embeddings from trained ST-GCN.
- Implement ProtoGCN-style prototype learning:
- Learn 3-5 class-specific prototypes per behavior in embedding space.
- Select real sequences closest to each prototype as cent...