HAN Humanoid Gesture Model v3
Overview
Hybrid CNN-LSTM model designed for humanoid gesture recognition.
The model classifies movement sequences into predefined gesture categories.
Model Type
CNN-LSTM Hybrid Network
Intended Use
- Gesture recognition
- Human-robot interaction
- Motion classification
- Robotics training simulation
Training Details
- Framework: PyTorch
- Epochs: 28
- Batch Size: 32
- Optimizer: AdamW
- Learning Rate: 0.0003
Input
- Motion capture sequences
- Skeletal joint coordinates
- Temporal movement frames
Output
12 gesture classification labels
Evaluation
- Classification Accuracy: 94%
- F1 Score: 0.92
Tags
humanoid, robotics, gesture-recognition, cnn-lstm, han-network
