File size: 2,448 Bytes
b2f457f
 
7d7434a
 
 
a7a8193
7d7434a
 
b2f457f
 
6b49d7b
 
7d7434a
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
6b49d7b
7d7434a
 
 
 
 
 
 
 
 
 
 
 
 
a7a8193
 
 
 
 
7d7434a
 
 
 
 
 
 
 
 
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
from transformers import pipeline
from PIL import Image, ImageDraw
from sentence_transformers import SentenceTransformer, util
import requests
from io import BytesIO
import gradio as gr

# Inisialisasi model deteksi dan pengkodean
checkpoint = "google/owlvit-base-patch32"
detector = pipeline(model=checkpoint, task="zero-shot-object-detection")
model = SentenceTransformer('clip-ViT-L-14')

def load_image(url):
    response = requests.get(url)
    image = Image.open(BytesIO(response.content)).convert("RGB")
    return image

def get_face_image(url):
    image = load_image(url)
    predictions = detector(
        image,
        candidate_labels=["human face"],
    )
    max_score = 0
    box_area = None
    for prediction in predictions:
        box = prediction["box"]
        label = prediction["label"]
        score = prediction["score"]
        if score > max_score:
            xmin, ymin, xmax, ymax = box.values()
            box_area = (xmin, ymin, xmax, ymax)
            max_score = score
        else:
            continue
    if box_area:
        draw = ImageDraw.Draw(image)
        draw.rectangle(box_area, outline="red", width=2)
        crop_img = image.crop(box_area)
        newsize = (256, 256)
        face_img = crop_img.resize(newsize)
        return face_img

def predict(url1, url2, inp_sim):
    face_image1 = get_face_image(url1)
    face_image2 = get_face_image(url2)

    if face_image1 is not None and face_image2 is not None:
        img_emb = model.encode([face_image1, face_image2])
        sim = util.cos_sim(img_emb[0], img_emb[1])
        if sim > inp_sim:
            return sim.item(), "SAME"
        else:
            return sim.item(), "DIFFERENT"
    else:
        return 0, "No face detected in one or both images"

description = "An application that can recognize if two faces belong to the same person or not"
title = "Facial Identity Recognition System"

interface = gr.Interface(fn=predict, 
                         inputs=[gr.Textbox(label="Image URL 1"),
                                 gr.Textbox(label="Image URL 2"),
                                 gr.Slider(0, 1, value=0.8, label="Similarity Percentage", info="Choose between 0 and 1")],
                         outputs=[gr.Number(label="Similarity"),
                                  gr.Textbox(label="Message")],
                         description=description,
                         title=title)

interface.launch(debug=True)