Spaces:
Sleeping
Sleeping
Download src/polygraph.py from turboryzen/VERIDIC: direct link, hf CLI and curl.
- Browser
- Download file 1.74 kB
-
https://huggingface.co/spaces/turboryzen/VERIDIC/resolve/main/src/polygraph.py
- Command line
-
hf download hf://spaces/turboryzen/VERIDIC/src/polygraph.py
-
curl -L -o polygraph.py https://huggingface.co/spaces/turboryzen/VERIDIC/resolve/main/src/polygraph.py
1.74 kB
| # src/polygraph.py | |
| import joblib, numpy as np, os | |
| _MODEL_PATH = os.path.join(os.path.dirname(__file__), "..", "models", "polygraph_xgb.pkl") | |
| _model = joblib.load(_MODEL_PATH) | |
| # Orden fijo que debe coincidir con el entrenamiento | |
| FEATURE_NAMES = [ | |
| "ear_mean", # apertura media de ojos (menor = mas tenso) | |
| "ear_std", # variabilidad de apertura | |
| "iris_ratio_mean", # tamano relativo del iris | |
| "iris_ratio_std", # variabilidad del iris | |
| "ear_asymmetry", # diferencia izquierdo-derecho EAR (asimetria ocular) | |
| "brow_asymmetry", # diferencia de elevacion de cejas | |
| "gaze_std", # inestabilidad de mirada (std del centro de iris) | |
| "head_roll_std", # variabilidad del giro lateral de cabeza | |
| "head_pitch_std", # variabilidad de cabeceo | |
| "head_yaw_std", # variabilidad de giro horizontal | |
| "hr_bpm", # frecuencia cardiaca estimada (rPPG) | |
| "hr_std", # variabilidad FC (HRV proxy) | |
| "pitch_mean", # tono medio de voz (F0) | |
| "pitch_std", # variabilidad de tono (jitter proxy) | |
| "energy_std", # variabilidad de energia vocal (shimmer proxy) | |
| "pause_ratio", # fraccion de silencio | |
| "speech_rate", # fraccion de habla activa | |
| "suppression_index", # FACS: supresion emocional (blendshapes asimetricos) | |
| "emotion_variance", # FACS: varianza inter-emocional (expresion inconsistente) | |
| "au_fear_peak", # FACS: pico de miedo/AU1+AU2+AU4 (microexpresiones) | |
| ] | |
| def compute_score(features: dict) -> float: | |
| """Devuelve probabilidad de engano [0-1].""" | |
| X = np.array([[features.get(k, 0.0) for k in FEATURE_NAMES]]) | |
| return float(_model.predict_proba(X)[0, 1]) | |