change the method of face detection to increase ac

Class/collect_data/face_reco_from_camera.py

@@ -9,7 +9,11 @@ import time
 from PIL import Image, ImageDraw, ImageFont
 
 # 1. Dlib 正向人脸检测器
-detector = dlib.get_frontal_face_detector()
+# detector = dlib.get_frontal_face_detector()
+
+# OpenCV DNN face detector
+detector = cv2.dnn.readNetFromCaffe("data/data_opencv/deploy.prototxt.txt",
+                                    "data/data_opencv/res10_300x300_ssd_iter_140000.caffemodel")
 
 # 2. Dlib 人脸 landmark 特征点检测器
 predictor = dlib.shape_predictor('data/data_dlib/shape_predictor_68_face_landmarks.dat')
@@ -53,7 +57,7 @@ class Face_Recognizer:
                     else:
                         features_someone_arr.append(csv_rd.iloc[i][j])
                 self.features_known_list.append(features_someone_arr)
-                self.name_known_list.append("Person_"+str(i+1))
+                self.name_known_list.append("Person_" + str(i + 1))
             self.name_known_cnt = len(self.name_known_list)
             print("Faces in Database：", len(self.features_known_list))
             return 1
@@ -74,6 +78,8 @@ class Face_Recognizer:
         dist = np.sqrt(np.sum(np.square(feature_1 - feature_2)))
         return dist
 
+    # TODO th
+
     # 更新 FPS
     def update_fps(self):
         now = time.time()
@@ -91,7 +97,7 @@ class Face_Recognizer:
 
     def draw_name(self, img_rd):
         # 在人脸框下面写人脸名字
-        font = ImageFont.truetype("simsun.ttc", 30)
+        font = ImageFont.truetype("simsun.ttc", 30, index=1)
         img = Image.fromarray(cv2.cvtColor(img_rd, cv2.COLOR_BGR2RGB))
         draw = ImageDraw.Draw(img)
         for i in range(self.faces_cnt):
@@ -104,7 +110,7 @@ class Face_Recognizer:
     def modify_name_camera_list(self):
         # TODO 数据库 ID
         # Default known name: person_1, person_2, person_3
-        self.name_known_list[0] ='唐麒'.encode('utf-8').decode()
+        self.name_known_list[0] = '唐麒'.encode('utf-8').decode()
         # self.name_known_list[1] ='李四'.encode('utf-8').decode()
         # self.name_known_list[2] ='xx'.encode('utf-8').decode()
         # self.name_known_list[3] ='xx'.encode('utf-8').decode()
@@ -116,7 +122,6 @@ class Face_Recognizer:
         if self.get_face_database():
             while stream.isOpened():
                 flag, img_rd = stream.read()
-                faces = detector(img_rd, 0)
                 kk = cv2.waitKey(1)
                 # 按下 q 键退出
                 if kk == ord('q'):
@@ -128,25 +133,49 @@ class Face_Recognizer:
                     self.pos_camera_list = []
                     self.name_camera_list = []
 
+                    (h, w) = img_rd.shape[:2]
+                    blob = cv2.dnn.blobFromImage(cv2.resize(img_rd, (300, 300)), 1.0,
+                                                 (300, 300), (104.0, 177.0, 123.0))
+                    detector.setInput(blob)
+                    faces = detector.forward()
+
                     # 2. 检测到人脸
-                    if len(faces) != 0:
+                    if faces.shape[2] != 0:
                         # 3. 获取当前捕获到的图像的所有人脸的特征，存储到 self.features_camera_list
-                        for i in range(len(faces)):
-                            shape = predictor(img_rd, faces[i])
+                        for i in range(0, faces.shape[2]):
+                            confidence = faces[0, 0, i, 2]
+
+                            # filter out weak detections by ensuring the `confidence` is
+                            # greater than the minimum confidence
+                            if confidence < 0.5:
+                                continue
+                            box = faces[0, 0, i, 3:7] * np.array([w, h, w, h])
+                            (startX, startY, endX, endY) = box.astype("int")
+                            rect = dlib.rectangle(startX, startY, endX, endY)
+                            shape = predictor(img_rd, rect)
                             self.features_camera_list.append(face_reco_model.compute_face_descriptor(img_rd, shape))
 
                         # 4. 遍历捕获到的图像中所有的人脸
-                        for k in range(len(faces)):
+                        for k in range(0, faces.shape[2]):
+                            # 计算矩形框大小
+                            confidence = faces[0, 0, k, 2]
+
+                            # filter out weak detections by ensuring the `confidence` is
+                            # greater than the minimum confidence
+                            if confidence < 0.5:
+                                continue
                             print("##### camera person", k + 1, "#####")
-                            # 让人名跟随在矩形框的下方
+                            # 让人名跟随在矩形框的上方
                             # 确定人名的位置坐标
                             # 先默认所有人不认识，是 unknown
                             # Set the default names of faces with "unknown"
                             self.name_camera_list.append("unknown")
 
                             # 每个捕获人脸的名字坐标
+                            box = faces[0, 0, k, 3:7] * np.array([w, h, w, h])
+                            (startX, startY, endX, endY) = box.astype("int")
                             self.pos_camera_list.append(tuple(
-                                [faces[k].left(), int(faces[k].bottom() + (faces[k].bottom() - faces[k].top()) / 4)]))
+                                [int(startX+5), int(startY - 30)]))
 
                             # 5. 对于某张人脸，遍历所有存储的人脸特征
                             e_distance_list = []
@@ -174,8 +203,10 @@ class Face_Recognizer:
                             # 矩形框
                             for kk, d in enumerate(faces):
                                 # 绘制矩形框
-                                cv2.rectangle(img_rd, tuple([d.left(), d.top()]), tuple([d.right(), d.bottom()]),
-                                              (0, 255, 255), 2)
+                                cv2.rectangle(img_rd, tuple([startX, startY]), tuple([endX, endY]),
+                                              (0, 255, 0), 2)
+                                cv2.rectangle(img_rd, tuple([startX, startY - 35]), tuple([endX, startY]),
+                                              (0, 255, 0), cv2.FILLED)
                             print('\n')
 
                         self.faces_cnt = len(faces)
@@ -210,4 +241,4 @@ def main():
 
 
 if __name__ == '__main__':
-    main()
+    main()

Class/collect_data/features_extraction_to_csv.py

@@ -4,13 +4,17 @@ import os
 import dlib
 from skimage import io
 import csv
+import cv2
 import numpy as np
 
 # 要读取人脸图像文件的路径
 path_images_from_camera = "data/data_faces_from_camera/"
 
 # 1. Dlib 正向人脸检测器
-detector = dlib.get_frontal_face_detector()
+# detector = dlib.get_frontal_face_detector()
+# OpenCV DNN face detector
+detector = cv2.dnn.readNetFromCaffe("data/data_opencv/deploy.prototxt.txt",
+                                    "data/data_opencv/res10_300x300_ssd_iter_140000.caffemodel")
 
 # 2. Dlib 人脸 landmark 特征点检测器
 predictor = dlib.shape_predictor('data/data_dlib/shape_predictor_68_face_landmarks.dat')
@@ -22,14 +26,21 @@ face_reco_model = dlib.face_recognition_model_v1("data/data_dlib/dlib_face_recog
 # 返回单张图像的 128D 特征
 def return_128d_features(path_img):
     img_rd = io.imread(path_img)
-    faces = detector(img_rd, 1)
+    (h, w) = img_rd.shape[:2]
+    blob = cv2.dnn.blobFromImage(cv2.resize(img_rd, (300, 300)), 1.0,
+                                 (300, 300), (104.0, 177.0, 123.0))
+    detector.setInput(blob)
+    faces = detector.forward()
 
     print("%-40s %-20s" % ("检测到人脸的图像 / Image with faces detected:", path_img), '\n')
 
     # 因为有可能截下来的人脸再去检测，检测不出来人脸了
     # 所以要确保是 检测到人脸的人脸图像 拿去算特征
-    if len(faces) != 0:
-        shape = predictor(img_rd, faces[0])
+    if faces.shape[2] != 0:
+        box = faces[0, 0, 0, 3:7] * np.array([w, h, w, h])
+        (startX, startY, endX, endY) = box.astype("int")
+        rect = dlib.rectangle(startX, startY, endX, endY)
+        shape = predictor(img_rd, rect)
         face_descriptor = face_reco_model.compute_face_descriptor(img_rd, shape)
     else:
         face_descriptor = 0

Class/collect_data/get_faces_from_camera.py

@@ -1,15 +1,20 @@
 # 进行人脸录入 / face register
 # 录入多张人脸 / support multi-faces
 
-import dlib         # 人脸处理的库 Dlib
-import numpy as np  # 数据处理的库 Numpy
-import cv2          # 图像处理的库 OpenCV
-import os           # 读写文件
-import shutil       # 读写文件
+import dlib
+import numpy as np
+import cv2
+import os
+import shutil  # 读写文件
 import time
+import face_recognition
 
 # Dlib 正向人脸检测器
-detector = dlib.get_frontal_face_detector()
+# detector = dlib.get_frontal_face_detector()
+
+# OpenCV DNN face detector
+detector = cv2.dnn.readNetFromCaffe("data/data_opencv/deploy.prototxt.txt",
+                                    "data/data_opencv/res10_300x300_ssd_iter_140000.caffemodel")
 
 
 class Face_Register:
@@ -17,9 +22,9 @@ class Face_Register:
         self.path_photos_from_camera = "data/data_faces_from_camera/"
         self.font = cv2.FONT_ITALIC
 
-        self.existing_faces_cnt = 0     # 已录入的人脸计数器
-        self.ss_cnt = 0                 # 录入 personX 人脸时图片计数器
-        self.faces_cnt = 0              # 录入人脸计数器
+        self.existing_faces_cnt = 0  # 已录入的人脸计数器
+        self.ss_cnt = 0  # 录入 personX 人脸时图片计数器
+        self.faces_cnt = 0  # 录入人脸计数器
 
         # 之后用来控制是否保存图像的 flag
         self.save_flag = 1
@@ -42,7 +47,7 @@ class Face_Register:
         # 删除之前存的人脸数据文件夹
         folders_rd = os.listdir(self.path_photos_from_camera)
         for i in range(len(folders_rd)):
-            shutil.rmtree(self.path_photos_from_camera+folders_rd[i])
+            shutil.rmtree(self.path_photos_from_camera + folders_rd[i])
         if os.path.isfile("data/features_all.csv"):
             os.remove("data/features_all.csv")
 
@@ -85,15 +90,21 @@ class Face_Register:
         # self.pre_work_mkdir()
 
         # 2. 删除 "/data/data_faces_from_camera" 中已有人脸图像文件
-        self.pre_work_del_old_face_folders()
+        # self.pre_work_del_old_face_folders()
 
         # 3. 检查 "/data/data_faces_from_camera" 中已有人脸文件
         self.check_existing_faces_cnt()
 
         while stream.isOpened():
-            flag, img_rd = stream.read()        # Get camera video stream
+            flag, img_rd = stream.read()  # Get camera video stream
+
             kk = cv2.waitKey(1)
-            faces = detector(img_rd, 0)         # Use dlib face detector
+
+            (h, w) = img_rd.shape[:2]
+            blob = cv2.dnn.blobFromImage(cv2.resize(img_rd, (300, 300)), 1.0,
+                                         (300, 300), (104.0, 177.0, 123.0))
+            detector.setInput(blob)
+            faces = detector.forward()
 
             # 4. 按下 'n' 新建存储人脸的文件夹
             if kk == ord('n'):
@@ -103,37 +114,50 @@ class Face_Register:
                 print('\n')
                 print("新建的人脸文件夹 / Create folders: ", current_face_dir)
 
-                self.ss_cnt = 0              # 将人脸计数器清零
-                self.press_n_flag = 1        # 已经按下 'n'
+                self.ss_cnt = 0  # 将人脸计数器清零
+                self.press_n_flag = 1  # 已经按下 'n'
 
             # 5. 检测到人脸
-            if len(faces) != 0:
+            if faces.shape[2] != 0:
                 # 矩形框
-                for k, d in enumerate(faces):
+                for i in range(0, faces.shape[2]):
                     # 计算矩形框大小
-                    height = (d.bottom() - d.top())
-                    width = (d.right() - d.left())
-                    hh = int(height/2)
-                    ww = int(width/2)
+                    confidence = faces[0, 0, i, 2]
+
+                    # filter out weak detections by ensuring the `confidence` is
+                    # greater than the minimum confidence
+                    if confidence < 0.5:
+                        continue
+
+                    # compute the (x, y)-coordinates of the bounding box for the
+                    # object
+                    box = faces[0, 0, i, 3:7] * np.array([w, h, w, h])
+                    (startX, startY, endX, endY) = box.astype("int")
+
+                    height = (endY - startY)
+                    width = (endX - startX)
+                    hh = int(height / 2)
+                    ww = int(width / 2)
 
                     # 6. 判断人脸矩形框是否超出 480x640
-                    if (d.right()+ww) > 640 or (d.bottom()+hh > 480) or (d.left()-ww < 0) or (d.top()-hh < 0):
+                    if (endX + ww) > 640 or (endY + hh > 480) or (startX - ww < 0) or (
+                            startY - hh < 0):
                         cv2.putText(img_rd, "OUT OF RANGE", (20, 300), self.font, 0.8, (0, 0, 255), 1, cv2.LINE_AA)
                         color_rectangle = (0, 0, 255)
                         save_flag = 0
                         if kk == ord('s'):
                             print("请调整位置 / Please adjust your position")
                     else:
-                        color_rectangle = (255, 255, 255)
+                        color_rectangle = (0, 255, 0)
                         save_flag = 1
 
                     cv2.rectangle(img_rd,
-                                  tuple([d.left() - ww, d.top() - hh]),
-                                  tuple([d.right() + ww, d.bottom() + hh]),
+                                  tuple([startX - ww, startY - hh]),
+                                  tuple([endX + ww, endY + hh]),
                                   color_rectangle, 2)
 
                     # 7. 根据人脸大小生成空的图像
-                    img_blank = np.zeros((int(height*2), width*2, 3), np.uint8)
+                    img_blank = np.zeros((int(height * 2), width * 2, 3), np.uint8)
 
                     if save_flag:
                         # 8. 按下 's' 保存摄像头中的人脸到本地
@@ -141,11 +165,12 @@ class Face_Register:
                             # 检查有没有先按'n'新建文件夹
                             if self.press_n_flag:
                                 self.ss_cnt += 1
-                                for ii in range(height*2):
-                                    for jj in range(width*2):
-                                        img_blank[ii][jj] = img_rd[d.top()-hh + ii][d.left()-ww + jj]
+                                for ii in range(height * 2):
+                                    for jj in range(width * 2):
+                                        img_blank[ii][jj] = img_rd[startY - hh + ii][startX - ww + jj]
                                 cv2.imwrite(current_face_dir + "/img_face_" + str(self.ss_cnt) + ".jpg", img_blank)
-                                print("写入本地 / Save into：", str(current_face_dir) + "/img_face_" + str(self.ss_cnt) + ".jpg")
+                                print("写入本地 / Save into：",
+                                      str(current_face_dir) + "/img_face_" + str(self.ss_cnt) + ".jpg")
                             else:
                                 print("请先按 'N' 来建文件夹, 按 'S' / Please press 'N' and press 'S'")
                 self.faces_cnt = len(faces)