diff --git a/comparerfmp b/comparerfmp
new file mode 100644
index 0000000000000000000000000000000000000000..c7913021ef26a0d8c09a7b1412192272efd7257d
--- /dev/null
+++ b/comparerfmp
@@ -0,0 +1,215 @@
+from deepface import DeepFace
+from deepface.commons import functions, realtime, distance as dst
+import matplotlib.pyplot as plt
+import cv2
+import glob
+import numpy as np
+import os
+import tensorflow as tf
+import mediapipe as mp
+from retinaface import RetinaFace
+import xlsxwriter
+import time
+
+#for deepface
+backends = ['opencv', 'ssd', 'Dlib', 'mtcnn', 'retinaface', 'mediapipe']
+backend = backends[3]
+
+#for mediapipe
+mp_face_detection = mp.solutions.face_detection
+mp_drawing = mp.solutions.drawing_utils
+mp_face_mesh = mp.solutions.face_mesh
+mp_drawing_styles = mp.solutions.drawing_styles
+fd = mp_face_detection.FaceDetection(min_detection_confidence=0.4, model_selection=1)
+fm = mp_face_mesh.FaceMesh(
+ static_image_mode=True,
+ max_num_faces=3,
+ refine_landmarks=True,
+ min_detection_confidence=0.5)
+
+
+#*******smaller sample
+# file = "/home/beyondem/Documents/comparing_models/*.jpg"
+# file = '/home/beyondem/Documents/comparing_models/retinadetected3/*.jpg'
+
+#*******bigger file
+file = "/home/beyondem/Documents/comparing_models/images/WIDER_train/**/*.png"
+pr_file = open("log.txt", "w")
+
+
+#****for retina
+# path1 = '/home/beyondem/Documents/comparing_models/retinadetected3'
+# path2 = '/home/beyondem/Documents/comparing_models/retinadetected3/data'
+
+#****for mp
+path1 = '/home/beyondem/Documents/comparing_models/withmp'
+path11 = '/home/beyondem/Documents/comparing_models/withmp/mp'
+path12 = '/home/beyondem/Documents/comparing_models/withmp/rf'
+path13 = '/home/beyondem/Documents/comparing_models/withmp/fm'
+path2 = '/home/beyondem/Documents/comparing_models/withmp/data'
+
+
+#get images
+images = glob.glob(file, recursive=True)
+print(len(images))
+
+
+#************ Create an new Excel file and add a worksheet.
+#>>>>>>>> all.xlsx is for all models compared together
+
+workbook = xlsxwriter.Workbook(os.path.join(path2,'all.xlsx'))
+worksheet = workbook.add_worksheet()
+
+# Widen the first column to make the text clearer.
+worksheet.set_column('A:A', 60)
+worksheet.set_column('B:B', 20)
+worksheet.set_column('C:C', 20)
+worksheet.set_column('D:D', 10)
+worksheet.set_column('E:E', 10)
+worksheet.set_column('F:F', 10)
+worksheet.set_column('G:G', 10)
+# Add a bold format to use to highlight cells.
+bold = workbook.add_format({'bold': True})
+
+
+worksheet.write(0,0, "file name")
+worksheet.write(0,1, "#boxes_retina")
+worksheet.write(0,2, "#boxes_mp")
+worksheet.write(0,3, "#time_retina")
+worksheet.write(0,4, "#time_mp")
+worksheet.write(0,5, "rf_average score")
+worksheet.write(0,6, "mp_average score")
+i = 1
+
+
+# For static images:
+IMAGE_FILES = []
+drawing_spec = mp_drawing.DrawingSpec(thickness=1, circle_radius=1)
+
+with fm as face_mesh:
+ with fd as face_detection:
+ for f in images:
+ print(f, file=pr_file)
+ image = cv2.imread(f)
+
+ #for mediapipe
+ start_mp = time.time()
+ imagergb = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
+ imagergb2 = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
+ obj_mp = face_detection.process(imagergb)
+ results = face_mesh.process(imagergb2)
+ # obj_mp = face_detection.process(cv2.cvtColor(image, cv2.COLOR_BGR2RGB))
+ # results = face_mesh.process(cv2.cvtColor(image, cv2.COLOR_BGR2RGB))
+ # print(obj_mp)
+ end_mp = time.time()
+
+
+ #for retinaface
+ start_r = time.time()
+ obj_ret = RetinaFace.detect_faces(f)
+ end_r = time.time()
+
+
+ # Write number of faces, with row/column notation.
+ worksheet.write(i, 0, os.path.join(os.path.basename(os.path.dirname(f)), os.path.basename(f)))
+ worksheet.write(i, 1, len(obj_ret))
+ # if type(obj_mp) is dict:
+ # if obj_mp.detections is not None:
+ # worksheet.write(i, 2, len(obj_mp.detections)*len(obj_mp.detections))
+ # else:
+ # worksheet.write(i, 2, 0)
+
+ #****************************************************************
+ rf_average_score = 0.0
+ #for drawing boxes using retinaface
+ if type(obj_ret) is dict:
+ for key in obj_ret.keys():
+ # print((obj_ret[key])['score'])
+ rf_average_score += (obj_ret[key])['score']
+ # print(key, file = pr_file)
+ identity = obj_ret[key]
+ facial_area = identity["facial_area"]
+ # worksheet.write(i, 2, facial_area[0])
+ # worksheet.write(i, 3, facial_area[1])
+ # worksheet.write(i, 4, facial_area[2])
+ # worksheet.write(i, 5, facial_area[3])
+ cv2.rectangle(image, (facial_area[2],facial_area[3]),(facial_area[0],facial_area[1]),(0,2,255),5)
+ rf_average_score = rf_average_score/ len(obj_ret)
+ # print(rf_average_score)
+ cv2.imwrite( os.path.join(path12 ,os.path.basename(f)), image)
+ cv2.waitKey(0)
+
+
+
+ #***********************************************************
+ #face mesh
+ annotated_image = imagergb2.copy()
+ if results.multi_face_landmarks is not None:
+ for face_landmarks in results.multi_face_landmarks:
+ print('face_landmarks:', face_landmarks,file=pr_file)
+ mp_drawing.draw_landmarks(
+ image=annotated_image,
+ landmark_list=face_landmarks,
+ connections=mp_face_mesh.FACEMESH_TESSELATION,
+ landmark_drawing_spec=None,
+ connection_drawing_spec=mp_drawing_styles
+ .get_default_face_mesh_tesselation_style())
+ mp_drawing.draw_landmarks(
+ image=annotated_image,
+ landmark_list=face_landmarks,
+ connections=mp_face_mesh.FACEMESH_CONTOURS,
+ landmark_drawing_spec=None,
+ connection_drawing_spec=mp_drawing_styles
+ .get_default_face_mesh_contours_style())
+ mp_drawing.draw_landmarks(
+ image=annotated_image,
+ landmark_list=face_landmarks,
+ connections=mp_face_mesh.FACEMESH_IRISES,
+ landmark_drawing_spec=None,
+ connection_drawing_spec=mp_drawing_styles
+ .get_default_face_mesh_iris_connections_style())
+ annotated_image = cv2.cvtColor(annotated_image, cv2.COLOR_RGB2BGR)
+ cv2.imwrite( os.path.join(path13 ,os.path.basename(f)), annotated_image)
+ #***********************************************************************************
+
+
+ #for displaying the images with the box
+ # Draw face detections of each face.
+ if not obj_mp.detections:
+ continue
+ image.flags.writeable = True
+ # image = cv2.cvtColor(image, cv2.COLOR_RGB2BGR)
+ counter = 0
+ mp_average_score = 0.0
+ # print(obj_mp.detections, file=pr_file)
+ if obj_mp.detections is not None:
+ # print((enumerate(obj_mp.detections)))
+ for counter, detection in enumerate(obj_mp.detections):
+ score = detection.score
+ box = detection.location_data.relative_bounding_box
+ # print(score,file=pr_file)
+ # print(box,file=pr_file)
+ counter += 1
+ mp_average_score = mp_average_score + score[0]
+
+
+ mp_drawing.draw_detection(imagergb, detection)
+ mp_average_score = mp_average_score/counter
+ print(counter, mp_average_score, file=pr_file)
+
+ imagergb= cv2.cvtColor(imagergb, cv2.COLOR_RGB2BGR)
+ cv2.imwrite( os.path.join(path11 ,os.path.basename(f)), imagergb)
+ cv2.waitKey(0)
+ # #**************************************************************
+ # print(counter)
+ worksheet.write(i, 2, counter)
+ worksheet.write(i, 3, end_r-start_r)
+ worksheet.write(i, 4, end_mp-start_mp)
+ worksheet.write(i, 5, rf_average_score)
+ worksheet.write(i, 6, mp_average_score)
+ i = i+1
+
+
+workbook.close()
+
+pr_file.close()