Haar分类器结合keras-facenet算法实现人脸检测分割及人脸识别考勤系统完整源码+项目说明.zip

# coding: utf-8 # ================================ # 人脸样本录入环节 # SEU-PR // R.YY & Z.HF & Z.X # ================================ import cv2 import PIL.Image from PIL import ImageTk from tkinter import * from tkinter import filedialog, messagebox import json import face_detection as FD from time import time TARGET_SIZE = (100, 100) howmany = 0 detected = [] # 摄像头捕获方式 camera = cv2.VideoCapture(0) # 摄像头 current_status = 'none' camera_looper = None def capture_camera(): global camera, camera_looper, detected, howmany if not camera.isOpened(): camera = cv2.VideoCapture(0) # 摄像头 status, frame = camera.read() if status: cv2.waitKey(10) (res, howmany, detected) = FD.detect_face_for_manager(frame) label_faceCount.config(text='当前人脸数：' + str(howmany)) res = cv2.resize(res, (640, int(640. / res.shape[1] * res.shape[0]))) res = cv2.cvtColor(res, cv2.COLOR_BGR2RGBA) res_pil = PIL.Image.fromarray(res) res_tk = ImageTk.PhotoImage(image=res_pil) display_area.imgtk = res_tk display_area.config(image=res_tk, width=res.shape[1], height=res.shape[0]) camera_looper = root.after(1, capture_camera) # 反复更新当前帧 # 本地文件方式 def capture_file(path: str): global detected, howmany if camera.isOpened() and camera_looper: root.after_cancel(camera_looper) camera.release() img = cv2.imread(path) (res, howmany, detected) = FD.detect_face_for_manager(img, int(img.shape[0] / 80) + 1) label_faceCount.config(text='当前人脸数：' + str(howmany)) res = cv2.resize(res, (640, int(640. / res.shape[1] * res.shape[0]))) res = cv2.cvtColor(res, cv2.COLOR_BGR2RGBA) res_pil = PIL.Image.fromarray(res) res_tk = ImageTk.PhotoImage(image=res_pil) display_area.imgtk = res_tk display_area.config(image=res_tk, width=res.shape[1], height=res.shape[0]) def ask_open_file(): path = filedialog.askopenfilename(title='打开图片...', filetypes=[('PNG Image', '*.png'), ('JPG Image', '*.jpg')]) capture_file(path) def save_sample(): global detected, howmany if howmany != 1: messagebox.showerror('错误', '当前人脸数不为1！') return face_gray = cv2.resize(detected[0], TARGET_SIZE) name = input_label.get() fp = open('./dataset/manager.json', 'r') json_content = json.load(fp) fp.close() total = int(json_content['total']) total += 1 json_content['total'] = total fp = open('./dataset/manager.json', 'w') json.dump(json_content, fp) fp.close() filename = str(total) + '-' + name save_path = './dataset/train/' + filename + '.png' cv2.imwrite(save_path, face_gray) messagebox.showinfo('录入成功', '录入成功！') root = Tk() root.title("人脸样本录入 - [SEU-PR]R.YY, Z.HF & Z.X") Label(root, {'text': '人脸样本录入 - [SEU-PR]', 'font': '宋体 16'}).pack() label_faceCount = Label(root, {'text': '当前人脸数：0', 'foreground': 'red'}) label_faceCount.pack() display_area = Label(root, {'width': 100, 'height': 30, 'text': '请选择一种录入方式开始...'}) display_area.pack() Button(root, {'text': '从摄像头录入', 'command': capture_camera}).pack({'side': LEFT}) Button(root, {'text': '从本地图片录入', 'command': ask_open_file}).pack({'side': LEFT}) Label(root, {'text': '请确保只有一张需要录入的人脸被框定！', 'foreground': 'red'}).pack({'side': LEFT}) Label(root, {'text': '请输入姓名：'}).pack({'side': LEFT}) input_label = Entry(root) input_label.pack({'side': LEFT}) Button(root, {'text': '保存', 'command': save_sample}).pack({'side': LEFT}) root.mainloop() camera.release()