AI-Solutions-KK/face_recognition

from huggingface_hub import hf_hub_download
import joblib
import numpy as np

REPO_ID = "AI-Solutions-KK/face_recognition"

svc_path = hf_hub_download(REPO_ID, "svc_model.pkl")
centroids_path = hf_hub_download(REPO_ID, "centroids.npy")
classes_path = hf_hub_download(REPO_ID, "classes.npy")

svc_model = joblib.load(svc_path)
centroids = np.load(centroids_path)
class_names = np.load(classes_path, allow_pickle=True)

print("Model loaded successfully. Classes:", len(class_names))

from huggingface_hub import hf_hub_download
import joblib, numpy as np, cv2, torch
from facenet_pytorch import InceptionResnetV1, MTCNN

REPO_ID = "AI-Solutions-KK/face_recognition"

# Load classifier + metadata
svc_path = hf_hub_download(REPO_ID, "svc_model.pkl")
classes_path = hf_hub_download(REPO_ID, "classes.npy")

obj = joblib.load(svc_path)
svc_model = obj["clf"]
normalizer = obj["norm"]
label_encoder = obj["le"]
class_names = np.load(classes_path, allow_pickle=True)

# Load FaceNet backbone + face detector
device = "cpu"
mtcnn = MTCNN(keep_all=False, device=device)
facenet = InceptionResnetV1(pretrained="vggface2").eval().to(device)

def get_embedding(img_path: str) -> np.ndarray:
    img_bgr = cv2.imread(img_path)
    if img_bgr is None:
        raise ValueError(f"Could not read image: {img_path}")
    img_rgb = cv2.cvtColor(img_bgr, cv2.COLOR_BGR2RGB)
    face = mtcnn(img_rgb)
    if face is None:
        raise ValueError("No face detected.")
    if face.dim() == 3:
        face = face.unsqueeze(0)
    with torch.no_grad():
        emb = facenet(face.to(device)).cpu().numpy()
    return emb

def predict_face(img_path: str):
    emb = get_embedding(img_path)
    emb_norm = normalizer.transform(emb)
    probs = svc_model.predict_proba(emb_norm)[0]
    idx = np.argmax(probs)
    label = label_encoder.inverse_transform([idx])[0]
    confidence = float(probs[idx])
    return label, confidence

# -------- RUN ----------
img_path = "test.jpg"
label, prob = predict_face(img_path)
print("Predicted Identity:", label)
print("Confidence Score:", prob)