update tracking

This commit is contained in:
Joseph 2025-01-21 13:35:04 +07:00
parent 07d28e24cd
commit b908185658
6 changed files with 189 additions and 77 deletions

2
DetectFace/.gitignore vendored Normal file
View File

@ -0,0 +1,2 @@
dataset
test

77
DetectFace/detect.py Normal file
View File

@ -0,0 +1,77 @@
import cv2
import face_recognition
import os
import numpy as np
import pickle
datasetPath = "dataset"
images = []
classNames = []
lisFileTrain = os.listdir(datasetPath)
for file in lisFileTrain:
currentImg = cv2.imread(f"{datasetPath}/{file}")
images.append(currentImg)
classNames.append(os.path.splitext(file)[0].split('_')[0])
print(len(images))
def encodeImgs(images, save_path="encodings.pkl"):
if os.path.exists(save_path):
print(f"Loading encodings from {save_path}...")
with open(save_path, "rb") as f:
return pickle.load(f)
encodeList = []
for i, img in enumerate(images):
print(i+1)
img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
encode = face_recognition.face_encodings(img)
if encode: # Check if encodings list is not empty
encodeList.append(encode[0])
else:
print("No face detected in an image. Skipping...")
os.remove(f"{datasetPath}/{lisFileTrain[i]}")
# Lưu encodeList vào file
print(f"Saving encodings to {save_path}...")
with open(save_path, "wb") as f:
pickle.dump(encodeList, f)
return encodeList
encodeListKnow = encodeImgs(images)
print("Load data success")
print(len(encodeListKnow))
cap = cv2.VideoCapture(0)
while True:
ret, frame = cap.read()
frameS = cv2.resize(frame, (0,0), None, fx=1, fy=1)
frameS = cv2.cvtColor(frameS, cv2.COLOR_BGR2RGB)
faceCurFrame = face_recognition.face_locations(frameS)
encodeCurFrame = face_recognition.face_encodings(frameS)
for encodeFace, faceLoc in zip(encodeCurFrame, faceCurFrame):
matches = face_recognition.compare_faces(encodeListKnow, encodeFace)
faceDis = face_recognition.face_distance(encodeListKnow, encodeFace)
print(faceDis)
matchIndex = np.argmin(faceDis)
if faceDis[matchIndex] < 0.3:
name = classNames[matchIndex].upper()
else:
name = "Unknow"
y1, x2, y2, x1 = faceLoc
y1, x2, y2, x1 = y1, x2, y2, x1
cv2.rectangle(frame, (x1,y1), (x2,y2), (0,255,0), 2)
cv2.putText(frame, name + f"({(1 - round(faceDis[matchIndex], 2))*100}%)", (x2, y2), cv2.FONT_HERSHEY_COMPLEX, 1, (255,255,255), 2)
cv2.imshow('Face decting', frame)
if cv2.waitKey(1) == ord("q"):
break
cap.release()
cv2.destroyAllWindows()

BIN
DetectFace/encodings.pkl Normal file

Binary file not shown.

BIN
DetectFace/listFiles.pkl Normal file

Binary file not shown.

View File

@ -1,66 +0,0 @@
import tensorflow as tf
import numpy as np
import cv2
import os
from tensorflow.keras import layers, models
from sklearn.model_selection import train_test_split
# 1. Load dữ liệu
def load_data(data_dir):
images = []
labels = []
for label, folder in enumerate(os.listdir(data_dir)):
folder_path = os.path.join(data_dir, folder)
for file in os.listdir(folder_path):
img_path = os.path.join(folder_path, file)
img = cv2.imread(img_path)
img = cv2.resize(img, (128, 128)) # Resize về kích thước cố định
images.append(img)
labels.append(label)
images = np.array(images) / 255.0 # Chuẩn hóa
labels = np.array(labels)
return images, labels
# Đường dẫn dữ liệu
data_dir = '/home/joseph/DetectFace/dataset'
images, labels = load_data(data_dir)
# Chia dữ liệu train/test
X_train, X_test, y_train, y_test = train_test_split(images, labels, test_size=0.2, random_state=42)
# 2. Tạo mô hình phát hiện khuôn mặt
model = models.Sequential([
layers.Conv2D(32, (3, 3), activation='relu', input_shape=(128, 128, 3)),
layers.MaxPooling2D((2, 2)),
layers.Conv2D(64, (3, 3), activation='relu'),
layers.MaxPooling2D((2, 2)),
layers.Conv2D(128, (3, 3), activation='relu'),
layers.Flatten(),
layers.Dense(128, activation='relu'),
layers.Dense(len(set(labels)), activation='softmax') # Số lớp tương ứng số nhãn
])
# Compile mô hình
model.compile(optimizer='adam',
loss='sparse_categorical_crossentropy',
metrics=['accuracy'])
# Huấn luyện mô hình
model.fit(X_train, y_train, epochs=10, validation_data=(X_test, y_test))
# 3. Lưu mô hình
model.save('face_detection_model.h5')
# 4. Sử dụng mô hình để dự đoán
def predict_face(image_path, model_path='face_detection_model.h5'):
model = tf.keras.models.load_model(model_path)
img = cv2.imread(image_path)
img_resized = cv2.resize(img, (128, 128)) / 255.0
img_resized = np.expand_dims(img_resized, axis=0)
prediction = model.predict(img_resized)
return np.argmax(prediction), np.max(prediction)
# Test với một ảnh
image_path = '/home/joseph/DetectFace/test/NGUYEN\ HOANG\ VI_check\ in_at_2024_09_05_11_31_24.png'
label, confidence = predict_face(image_path)
print(f"Detected label: {label}, Confidence: {confidence:.2f}")

View File

@ -6,6 +6,9 @@ import pyautogui
from pyzbar import pyzbar
from datetime import datetime
import requests
import face_recognition
import numpy as np
import pickle
# Khởi tạo danh sách rỗng để lưu trữ thông tin người dùng
user_data = []
history = []
@ -193,6 +196,70 @@ def process_qr_code(frame):
cv.destroyWindow(WINDOW_QR_CODE)
return frame
# Hàm để xử lý quá trình quét mã QR code
def process_face_detect(text, frame):
if text.endswith("\n\n"):
file_name = ""
status = ""
id_log = 0
if text not in [user["name"] for user in user_data]:
print(f"{text} đã check in lúc {datetime.now()}")
status += "check in"
file_name+=text.split('\n')[0]+"_"+f"{status}_at_{datetime.now().strftime("%Y_%m_%d_%H_%M_%S")}.png"
# screenshot_window(file_name)
res = check_in(text, frame, text)
id_log = res.get('data').get('id')
else:
print(f"{text} đã check out lúc {datetime.now()}")
status += "check out"
file_name+=text.split('\n')[0]+"_"+f"{status}_at_{datetime.now().strftime("%Y_%m_%d_%H_%M_%S")}.png"
# screenshot_window(file_name)
res = check_out(text, frame, text)
id_log = res.get('data').get('id')
cv.namedWindow(WINDOW_QR_CODE, cv.WINDOW_NORMAL)
cv.resizeWindow(WINDOW_QR_CODE, screen_width, screen_height)
cv.imshow(WINDOW_QR_CODE, frame)
cv.moveWindow(WINDOW_QR_CODE, 10, 10)
cv.waitKey(5000) # Chờ 5 giây
screenshot_window(file_name)
send_image(id_log, file_name)
cv.destroyWindow(WINDOW_QR_CODE)
else:
display_text(frame, f"QR invalid", (25, 25), 0.7, (6, 6, 255), 2)
display_text(frame, f"Failed", (25, 50), 0.7, (6, 6, 255), 2)
speak("Failed")
cv.namedWindow(WINDOW_QR_CODE, cv.WINDOW_NORMAL)
cv.resizeWindow(WINDOW_QR_CODE, screen_width, screen_height)
cv.imshow(WINDOW_QR_CODE, frame)
cv.moveWindow(WINDOW_QR_CODE, 10, 10)
cv.waitKey(2000)
cv.destroyWindow(WINDOW_QR_CODE)
return frame
datasetPath = "../DetectFace/dataset"
listFilesPath = '../DetectFace/listFiles.pkl'
images = []
classNames = []
lisFileTrain = []
if os.path.exists(listFilesPath):
with open(listFilesPath, 'rb') as f:
lisFileTrain = pickle.load(f)
else:
lisFileTrain = os.listdir(datasetPath)
with open(listFilesPath, 'wb') as f:
pickle.dump(lisFileTrain, f)
for file in lisFileTrain:
classNames.append(os.path.splitext(file)[0].split('_')[0])
def encodeImgs(save_path="../DetectFace/encodings.pkl"):
if os.path.exists(save_path):
print(f"Loading encodings from {save_path}...")
with open(save_path, "rb") as f:
return pickle.load(f)
encodeListKnow = encodeImgs()
print("Load data success")
# Khởi tạo camera
def main():
cap = cv.VideoCapture(0)
@ -203,20 +270,52 @@ def main():
ret, frame = cap.read()
if not ret:
break
frameS = cv.resize(frame, (0,0), None, fx=1, fy=1)
frameS = cv.cvtColor(frameS, cv.COLOR_BGR2RGB)
faceCurFrame = face_recognition.face_locations(frameS)
encodeCurFrame = face_recognition.face_encodings(frameS)
frame = process_qr_code(frame)
for encodeFace, faceLoc in zip(encodeCurFrame, faceCurFrame):
matches = face_recognition.compare_faces(encodeListKnow, encodeFace)
faceDis = face_recognition.face_distance(encodeListKnow, encodeFace)
print(faceDis)
matchIndex = np.argmin(faceDis)
if faceDis[matchIndex] < 0.3:
name = classNames[matchIndex].upper()
process_face_detect(f"{name}\n{"Staff"}\n\n", frame)
else:
name = "Unknow"
display_text(frame, f"Face not found - use QRcode", (20, 55), 0.7, (6, 6, 255), 2)
y1, x2, y2, x1 = faceLoc
y1, x2, y2, x1 = y1, x2, y2, x1
cv.rectangle(frame, (x1,y1), (x2,y2), (0,255,0), 2)
cv.putText(frame, name + f"({(1 - round(faceDis[matchIndex], 2))*100}%)", (20, 25), cv.FONT_HERSHEY_COMPLEX, 1, (255,255,255), 2)
# Convert the frame to grayscale
gray_frame = cv.cvtColor(frame, cv.COLOR_BGR2GRAY)
# gray_frame = cv.cvtColor(frame, cv.COLOR_BGR2GRAY)
# Detect faces in the frame
faces = face_cascade.detectMultiScale(gray_frame, scaleFactor=1.1, minNeighbors=25, minSize=(30, 30))
# # Detect faces in the frame
# faces = face_cascade.detectMultiScale(gray_frame, scaleFactor=1.1, minNeighbors=25, minSize=(30, 30))
# Draw rectangles around the faces
if len(faces) == 1:
for (x, y, w, h) in faces:
cv.rectangle(frame, (x, y), (x+w, y+h), (0, 255, 0), 2)
display_text(frame, f"Face detected", (430, 25), 0.7, (0, 255, 0), 2)
frame = process_qr_code(frame)
else:
display_text(frame, f"Face not found", (430, 25), 0.7, (6, 6, 255), 2)
# # Draw rectangles around the faces
# if len(faces) == 1:
# for (x, y, w, h) in faces:
# cv.rectangle(frame, (x, y), (x+w, y+h), (0, 255, 0), 2)
# display_text(frame, f"Face detected", (430, 25), 0.7, (0, 255, 0), 2)
# frame = process_qr_code(frame)
# else:
# display_text(frame, f"Face not found", (430, 25), 0.7, (6, 6, 255), 2)
cv.imshow(WINDOW_TRACKING, frame)
cv.moveWindow(WINDOW_TRACKING, 10, 10)
if cv.waitKey(1) == ord('q'):