Humanoid Gesture Classifier
Objective
Mengklasifikasikan jenis gesture humanoid berdasarkan parameter sudut sendi dan kecepatan tangan.
Problem Type
Multiclass Classification
Input Features
- shoulder_angle_deg
- elbow_angle_deg
- wrist_rotation_deg
- hand_velocity
- gesture_duration_sec
Output
- gesture_label (wave / point / grab / idle)
Architecture
- Input Normalization
- Dense(128) + ReLU
- Dense(64) + ReLU
- Dense(4) + Softmax
Loss Function
Categorical Crossentropy
Optimizer
Adam
Metrics
Accuracy F1 Score
Use Case
- Human-robot interaction
- Gesture-based control system