import cv2
import matplotlib.pyplot as plt
import streamlit as st
from deepface import DeepFace
import mediapipe
import os
import tempfile

backends = [
    'opencv', 
    'ssd', 
    'dlib', 
    'mtcnn', 
    'fastmtcnn',
    'retinaface', 
    'mediapipe',
    'yolov8',
    'yunet',
    'centerface',
]
metrics = ["cosine", "euclidean", "euclidean_l2"]
models = [
    "VGG-Face", 
    "Facenet", 
    "Facenet512", 
    "OpenFace", 
    "DeepFace", 
    "DeepID", 
    "ArcFace", 
    "Dlib", 
    "SFace",
    "GhostFaceNet",
]

def verify(img1, img2, model_name, backend, metric):
    # Save the uploaded images to temporary files
    with tempfile.NamedTemporaryFile(delete=False, suffix=".jpg") as temp_img1:
        temp_img1.write(img1.read())
        temp_img1_path = temp_img1.name

    with tempfile.NamedTemporaryFile(delete=False, suffix=".jpg") as temp_img2:
        temp_img2.write(img2.read())
        temp_img2_path = temp_img2.name

    img1p = cv2.imread(temp_img1_path)
    img2p = cv2.imread(temp_img2_path)

    img1p = cv2.cvtColor(img1p, cv2.COLOR_BGR2RGB)
    img2p = cv2.cvtColor(img2p, cv2.COLOR_BGR2RGB)

    face_detect = mediapipe.solutions.face_detection
    face_detector = face_detect.FaceDetection(min_detection_confidence=0.6)

    width1, height1 = img1p.shape[1], img1p.shape[0]
    width2, height2 = img2p.shape[1], img2p.shape[0]

    result1 = face_detector.process(img1p)
    result2 = face_detector.process(img2p)
    if result1.detections is not None:
      for face in result1.detections:
          if face.score[0] > 0.80:
              bounding_box = face.location_data.relative_bounding_box
              x = int(bounding_box.xmin * width1)
              w = int(bounding_box.width * width1)
              y = int(bounding_box.ymin * height1)
              h = int(bounding_box.height * height1)
              cv2.rectangle(img1p, (x, y), (x+w, y+h), color=(126, 133, 128), thickness=10)
    if result2.detections is not None:
      for face in result2.detections:
          if face.score[0] > 0.80:
              bounding_box = face.location_data.relative_bounding_box
              x = int(bounding_box.xmin * width2)
              w = int(bounding_box.width * width2)
              y = int(bounding_box.ymin * height2)
              h = int(bounding_box.height * height2)
              cv2.rectangle(img2p, (x, y), (x+w, y+h), color=(126, 133, 128), thickness=10)

    st.image([img1p, img2p], caption=["Image 1", "Image 2"], width=200)

    face = DeepFace.verify(img1p, img2p, model_name=model_name, detector_backend=backend, distance_metric=metric)
    verification = face["verified"]

    if verification:
        st.write("Matched")
    else:
        st.write("Not Matched")

# Streamlit app
def main():
    st.title("Face Verification App")
    tab_selection = st.sidebar.selectbox("Select Functionality", ["Face Verification", "Face Recognition", "Celebrity Lookalike", "Age and Emotions Detection"])

    if tab_selection == "Face Verification":
      st.header("Face Verification")
      model_name = st.selectbox("Select Model", models)
      backend = st.selectbox("Select Backend", backends)
      metric = st.selectbox("Select Metric", metrics)

      uploaded_img1 = st.file_uploader("Upload Image 1", type=["jpg", "png"])
      uploaded_img2 = st.file_uploader("Upload Image 2", type=["jpg", "png"])

      if uploaded_img1 and uploaded_img2:
          if st.button("Verify Faces"):
              verify(uploaded_img1, uploaded_img2, model_name, backend, metric)

# Run the app
if __name__ == "__main__":
    main()