facematching/webcam.py at master · Fluxmux/facematching · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
import cv2
import torchvision.transforms as transforms
from PIL import Image
import torch
import torch.nn.functional as F
from torch.utils.data import DataLoader
import torchvision.datasets as dset
import torchvision
import datetime

# user defined imports
from model import *
from helper_functions import *
from config import *
from dataset_helper import *
from loss_function import *
from yoloface import *

def load_dataset():
    folder_dataset_webcam = dset.ImageFolder(root=Config.webcam_dir)
    siamese_dataset_webcam = SiameseNetworkDataset(imageFolderDataset=folder_dataset_webcam,
                                            transform=transforms.Compose([transforms.Resize((100,100)),
                                                                          transforms.ToTensor()
                                                                          ])
                                           ,should_invert=False)


    webcam_dataloader = DataLoader(siamese_dataset_webcam,
                                 batch_size=1,
                                 shuffle=True)
    return webcam_dataloader

def crop_img(frame, left, top, width, height, filename):
    gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
    crop_gray = gray[top:top+height, left:left+width]
    pil = Image.fromarray(crop_gray)
    pil = transforms.functional.resize(pil, 100)
    pil.save(Config.webcam_dir + "s41/" + filename + ".pgm")

def test(net, img0, img1):
    device = torch.device("cuda:7" if torch.cuda.is_available() else "cpu")
    net.eval()
    with torch.no_grad():
        if isinstance(net, (SiameseNetwork,SiameseResNet)):
            output1, output2 = net(img0.to(device), img1.to(device))
            euclidean_distance = F.pairwise_distance(output1, output2)
            concatenated = torch.cat((img0,img1),0)
            imshow(torchvision.utils.make_grid(concatenated),'Score: {:.2f}'.format(euclidean_distance.item()))
            return euclidean_distance.item()

        elif isinstance(net, (SiameseNetworkAbs, SiameseResNetConcat)):
            output = net(img0.to(device), img1.to(device))
            probability = torch.sigmoid(output)
            concatenated = torch.cat((img0,img1),0)
            imshow(torchvision.utils.make_grid(concatenated),'Score: {:.2f}'.format(probability.item()))
            return probability.item()


def main():
    device = torch.device("cuda:7" if torch.cuda.is_available() else "cpu")
    Config.run_type = "webcam"

    net = torch.load("models/model.pt").to(device)    # selects neural net

    cam = cv2.VideoCapture(0)   # selects camera

    if not(cam.isOpened()):
        cam.open()

    print("press 'q' to quit")
    print("press 'a' and 'b' to capture image A and B")
    print("press 'c' to compare")
    img0 = img1 = None
    text = None
    color = None
    while(True):
        # Capture frame-by-frame
        ret, frame = cam.read()
        frame = cv2.flip(frame, 1)
        cv2.putText(frame,text,(0,30),cv2.FONT_HERSHEY_SIMPLEX,1,color,2)
        cv2.imshow('cam feed', frame)

        key = cv2.waitKey(1)

        if key & 0xFF == ord('q'):  # close camera
            cam.release()
            cv2.destroyAllWindows()
            break
        elif key & 0xFF == ord('a'):    # capture single frame
            filename = "1"
#            create_img(frame, filename)
            left, top, width, height = detect_faces(frame)         # excecute face detection and return coordinates
            crop_img(frame, left, top, width, height, filename)
            print("captured", filename)
        elif key & 0xFF == ord('b'):    # capture single frame
            filename = "2"
            left, top, width, height = detect_faces(frame)         # excecute face detection and return coordinates
            crop_img(frame, left, top, width, height, filename)
            print("captured", filename)
        elif key & 0xFF == ord('c'):    # compare the two images
            webcam_dataloader = load_dataset()
            print("testing...")
            for i, train_data in enumerate(webcam_dataloader, 0):
                img0, img1, _ = train_data
                img0 = img0.to(device)
                img1 = img1.to(device)
                score = test(net, img0, img1)
                print("\nscore: " + str(score))
                if isinstance(net, (SiameseNetwork,SiameseResNet)):
                    if score < Config.threshold:
                        text = "Same"
                        color = (0,255,0)
                        print("This is the same person!\n")
                    else:
                        text = "Different"
                        color = (0,0,255)
                        print("This is a different face!\n")
                elif isinstance(net, (SiameseNetworkAbs, SiameseResNetConcat)):
                    if score < 0.5:
                        text = "Same"
                        color = (0,255,0)
                        print("This is the same person!\n")
                    else:
                        text = "Different"
                        color = (0,0,255)
                        print("This is a different face!\n")
                #break

if __name__ == '__main__':
    main()