app/Hackathon_setup/face_recognition.py

import numpy as np
import cv2
from matplotlib import pyplot as plt
import torch
import torch.nn.functional as F
# In the below line,remove '.' while working on your local system. However Make sure that '.' is present before face_recognition_model while uploading to the server, Do not remove it.
from .face_recognition_model import *
from .face_recognition_model import Siamese
from PIL import Image
import base64
import io
import os
import joblib
import pickle
# Add more imports if required



###########################################################################################################################################
#         Caution: Don't change any of the filenames, function names and definitions                                                      #
#        Always use the current_path + file_name for refering any files, without it we cannot access files on the server                  # 
###########################################################################################################################################

# Current_path stores absolute path of the file from where it runs. 
current_path = os.path.dirname(os.path.abspath(__file__))

#1) The below function is used to detect faces in the given image.
#2) It returns only one image which has maximum area out of all the detected faces in the photo.
#3) If no face is detected,then it returns zero(0).

def detected_face(image):
    eye_haar = current_path + '/haarcascade_eye.xml'
    face_haar = current_path + '/haarcascade_frontalface_default.xml'
    face_cascade = cv2.CascadeClassifier(face_haar)
    eye_cascade = cv2.CascadeClassifier(eye_haar)
    gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
    faces = face_cascade.detectMultiScale(gray, 1.3, 5)
    face_areas=[]
    images = []
    required_image=0
    for i, (x,y,w,h) in enumerate(faces):
        face_cropped = gray[y:y+h, x:x+w]
        face_areas.append(w*h)
        images.append(face_cropped)
        required_image = images[np.argmax(face_areas)]
        required_image = Image.fromarray(required_image)
    return required_image


#1) Images captured from mobile is passed as parameter to the below function in the API call. It returns the similarity measure between given images.
#2) The image is passed to the function in base64 encoding, Code for decoding the image is provided within the function.
#3) Define an object to your siamese network here in the function and load the weight from the trained network, set it in evaluation mode.
#4) Get the features for both the faces from the network and return the similarity measure, Euclidean,cosine etc can be it. But choose the Relevant measure.
#5) For loading your model use the current_path+'your model file name', anyhow detailed example is given in comments to the function 
#Caution: Don't change the definition or function name; for loading the model use the current_path for path example is given in comments to the function
def get_similarity(img1, img2):
    device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
    
    det_img1 = detected_face(img1)
    det_img2 = detected_face(img2)
    if(det_img1 == 0 or det_img2 == 0):
        det_img1 = Image.fromarray(cv2.cvtColor(img1, cv2.COLOR_BGR2GRAY))
        det_img2 = Image.fromarray(cv2.cvtColor(img2, cv2.COLOR_BGR2GRAY))
    face1 = trnscm(det_img1).unsqueeze(0)
    face2 = trnscm(det_img2).unsqueeze(0)
    ##########################################################################################
    ##Example for loading a model using weight state dictionary:                            ##
    ## feature_net = light_cnn() #Example Network                                           ##
    ## model = torch.load(current_path + '/siamese_model.t7', map_location=device)          ##
    ## feature_net.load_state_dict(model['net_dict'])                                       ##
    ##                                                                                      ##
    ##current_path + '/<network_definition>' is path of the saved model if present in       ##
    ##the same path as this file, we recommend to put in the same directory                 ##
    ##########################################################################################
    ##########################################################################################
    
    # YOUR CODE HERE, load the model
    device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
    myModel = Siamese().to(device)
    BASE_DIR = os.path.dirname(os.path.abspath(__file__))
    ckpt_path = os.path.join(BASE_DIR, "siamese_model.t7")
    ckpt = torch.load(ckpt_path, map_location=device)
    myModel.load_state_dict(ckpt['net_dict'])
    myModel.eval()

    # Forward pass
    with torch.no_grad():
        output1, output2 = myModel(face1, face2)
        euclidean_distance = F.pairwise_distance(output1, output2)

    
    # YOUR CODE HERE, return similarity measure using your model
    euclidean_distance = F.pairwise_distance(output1, output2)
    similarity = 1 / (1 + euclidean_distance.item())


    return similarity
    
#1) Image captured from mobile is passed as parameter to this function in the API call, It returns the face class in the string form ex: "Person1"
#2) The image is passed to the function in base64 encoding, Code to decode the image provided within the function
#3) Define an object to your network here in the function and load the weight from the trained network, set it in evaluation mode
#4) Perform necessary transformations to the input(detected face using the above function).
#5) Along with the siamese, you need the classifier as well, which is to be finetuned with the faces that you are training
##Caution: Don't change the definition or function name; for loading the model use the current_path for path example is given in comments to the function
def get_face_class(img1):
    device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
    BASE_DIR = os.path.dirname(os.path.abspath(__file__))

    # 1 Load the Decision Tree classifier
    # clf_path = os.path.join(BASE_DIR, "decision_tree_model.sav")
    clf_path = os.path.join(BASE_DIR, "SVC_3.sav")
    clf = joblib.load(clf_path)

    scaler_path = os.path.join(BASE_DIR, "scaler.joblib")
    scaler = joblib.load(scaler_path)

    # 2 Load the Siamese feature extractor
    myModel = Siamese().to(device)
    ckpt_path = os.path.join(BASE_DIR, "siamese_model_1.t7")
    ckpt = torch.load(ckpt_path, map_location=device)
    myModel.load_state_dict(ckpt['net_dict'])
    myModel.eval()
    # myModel = myModel.float()
    # 3 Face detection (if available)
    # det_img1 = detected_face(img1)   # returns cropped face or 0 if not detected
    # if det_img1 == 0:
    #     # fallback: use original image
    #     det_img1 = Image.fromarray(cv2.cvtColor(img1, cv2.COLOR_BGR2GRAY))

    # 4 Transform the face
    img_tensor = transform1(img1).unsqueeze(0).to(device)

    # 5 Extract embeddings
    with torch.no_grad():
        embedding = myModel.forward_once(img_tensor)
        embedding = embedding.view(embedding.size(0), -1).cpu().numpy()  # shape (1, embedding_dim)

    # 6 Predict class using Decision Tree
    pred_label = clf.predict(scaler.transform(embedding))[0]


    # --- Predict ---
    # scaled_emb = scaler.transform(embedding)
    # probs = clf.predict_proba(scaled_emb)
    # pred_label = np.argmax(probs)
    # confidence = probs[0, pred_label]



    # 7 Optional: return class name (if available)
    # If you have the dataset available:
    # class_names = finalClassifierDset.classes
    # return class_names[pred_label]
    # class_names = ['Aayush', 'Aditya', 'Vikram']
    # return class_names[pred_label] + " " + str(pred_label)
    class_names = ['Aayush', 'Aditya', 'Vikram']
    return f"{class_names[pred_label]} {pred_label} {embedding}"


# def get_face_class(img1):
#
#     device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
#     BASE_DIR = os.path.dirname(os.path.abspath(__file__))
#
#     # 1 Load the Decision Tree classifier
#     # clf_path = os.path.join(BASE_DIR, "decision_tree_model.sav")
#     clf_path = os.path.join(BASE_DIR, "SVC_3.sav")
#     clf = joblib.load(clf_path)
#
#     scaler_path = os.path.join(BASE_DIR, "scaler.joblib")
#     scaler = joblib.load(scaler_path)
#
#     # 2 Load the Siamese feature extractor
#     myModel = Siamese().to(device)
#     ckpt_path = os.path.join(BASE_DIR, "siamese_model.t7")
#     ckpt = torch.load(ckpt_path, map_location=device)
#
#     myModel.load_state_dict(ckpt['net_dict'])
#     myModel.eval()
#     myModel = myModel.float()
#
#     img_tensor = transform1(img1).unsqueeze(0).to(device).float()
#
#     with torch.no_grad():
#         embedding = myModel.forward_once(img_tensor)
#     embedding = embedding.view(embedding.size(0), -1).cpu().numpy()  # shape (1, embedding_dim)
#     pred_label = clf.predict(scaler.transform(embedding))[0]
#
#     class_names = ['Aayush', 'Aditya', 'Vikram']
#     return f"{class_names[pred_label]} {pred_label} {embedding}"


# def get_face_class(img1):
#     """
#     img1: BGR image as numpy array (from cv2) OR path string accepted by detected_face.
#     Returns: "Name label_index" or debug info.
#     """
#
#     device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
#     BASE_DIR = os.path.dirname(os.path.abspath(__file__))
#
#     # 1) Load classifier + scaler
#     clf_path = os.path.join(BASE_DIR, "logistic_regression_5.sav")
#     scaler_path = os.path.join(BASE_DIR, "standar_scaler.sav")
#     clf = joblib.load(clf_path)
#     scaler = joblib.load(scaler_path)
#
#     # 2) Load Siamese feature extractor
#     myModel = Siamese().to(device)
#     ckpt_path = os.path.join(BASE_DIR, "siamese_model.t7")
#     ckpt = torch.load(ckpt_path, map_location=device)
#     myModel.load_state_dict(ckpt['net_dict'])
#     myModel.eval()
#
#     # 3) Face detection & crop
#     det_img1 = detected_face(img1)   # your function: should return cropped face (preferably PIL.Image or np.uint8)
#     if det_img1 == 0:
#         # fallback: convert original to grayscale PIL
#         if isinstance(img1, str):
#             pil_img = Image.open(img1).convert("L")
#         else:
#             # img1 assumed BGR numpy (cv2)
#             gray = cv2.cvtColor(img1, cv2.COLOR_BGR2GRAY)
#             pil_img = Image.fromarray(gray)
#         det_img1 = pil_img
#
#     # Ensure det_img1 is a PIL Image in mode 'L' (single channel). Convert if needed.
#     if isinstance(det_img1, np.ndarray):
#         # if it's color BGR -> convert to gray
#         if det_img1.ndim == 3 and det_img1.shape[2] == 3:
#             det_img1 = cv2.cvtColor(det_img1, cv2.COLOR_BGR2GRAY)
#         det_img1 = Image.fromarray(det_img1)
#     det_img1 = det_img1.convert("L")  # enforce single-channel
#
#     # 4) Transform the face: trnscm must be the exact same transform used when creating embeddings
#     img_tensor = trnscm(det_img1).unsqueeze(0)     # shape: (1, C, H, W)
#     img_tensor = img_tensor.to(device)             # <--- IMPORTANT: move to device!
#
#     # 5) Extract embeddings
#     with torch.no_grad():
#         embedding_t = myModel.forward_once(img_tensor)   # tensor on device
#         embedding_t = embedding_t.view(embedding_t.size(0), -1)
#         embedding = embedding_t.cpu().numpy()           # shape (1, embedding_dim)
#
#     # Debug prints (uncomment if needed)
#     # print("embedding shape:", embedding.shape)
#     # print("embedding min/max:", embedding.min(), embedding.max())
#     # print("embedding mean/std:", embedding.mean(), embedding.std())
#
#     # 6) Check for NaNs / inf
#     if np.isnan(embedding).any() or np.isinf(embedding).any():
#         return "ERROR: embedding contains NaN or inf"
#
#     # 7) Scale + predict
#     try:
#         scaled = scaler.transform(embedding)   # ensure scaler expects shape (1, D)
#     except Exception as e:
#         return f"Scaler transform error: {e}"
#
#     try:
#         pred_label = clf.predict(scaled)[0]
#     except Exception as e:
#         return f"Classifier predict error: {e}"
#
#     # 8) Optional: probabilities (if classifier supports it)
#     confidence = None
#     if hasattr(clf, "predict_proba"):
#         try:
#             probs = clf.predict_proba(scaled)
#             confidence = float(probs.max())
#         except Exception:
#             confidence = None
#
#     # 9) Map to class names
#     class_names = ['Aayush', 'Aditya', 'Vikram']  # replace with your saved names or load from file
#     name = class_names[pred_label] if pred_label < len(class_names) else str(pred_label)
#
#     if confidence is not None:
#         return f"{name} {pred_label} (conf={confidence:.3f})"
#     else:
#         return f"{name} {pred_label}"