File size: 5,789 Bytes
36f0d8c |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 |
import gradio as gr
import requests
import datadog_api_client
from PIL import Image
def check_liveness(frame):
url = "http://127.0.0.1:8080/check_liveness"
file = {'file': open(frame, 'rb')}
r = requests.post(url=url, files=file)
result = r.json().get('face_state').get('result')
html = None
faces = None
if r.json().get('face_state').get('is_not_front') is not None:
liveness_score = r.json().get('face_state').get('liveness_score')
eye_closed = r.json().get('face_state').get('eye_closed')
is_boundary_face = r.json().get('face_state').get('is_boundary_face')
is_not_front = r.json().get('face_state').get('is_not_front')
is_occluded = r.json().get('face_state').get('is_occluded')
is_small = r.json().get('face_state').get('is_small')
luminance = r.json().get('face_state').get('luminance')
mouth_opened = r.json().get('face_state').get('mouth_opened')
quality = r.json().get('face_state').get('quality')
html = ("<table>"
"<tr>"
"<th>Face State</th>"
"<th>Value</th>"
"</tr>"
"<tr>"
"<td>Result</td>"
"<td>{result}</td>"
"</tr>"
"<tr>"
"<td>Liveness Score</td>"
"<td>{liveness_score}</td>"
"</tr>"
"<tr>"
"<td>Quality</td>"
"<td>{quality}</td>"
"</tr>"
"<tr>"
"<td>Luminance</td>"
"<td>{luminance}</td>"
"</tr>"
"<tr>"
"<td>Is Small</td>"
"<td>{is_small}</td>"
"</tr>"
"<tr>"
"<td>Is Boundary</td>"
"<td>{is_boundary_face}</td>"
"</tr>"
"<tr>"
"<td>Is Not Front</td>"
"<td>{is_not_front}</td>"
"</tr>"
"<tr>"
"<td>Face Occluded</td>"
"<td>{is_occluded}</td>"
"</tr>"
"<tr>"
"<td>Eye Closed</td>"
"<td>{eye_closed}</td>"
"</tr>"
"<tr>"
"<td>Mouth Opened</td>"
"<td>{mouth_opened}</td>"
"</tr>"
"</table>".format(liveness_score=liveness_score, quality=quality, luminance=luminance, is_small=is_small, is_boundary_face=is_boundary_face,
is_not_front=is_not_front, is_occluded=is_occluded, eye_closed=eye_closed, mouth_opened=mouth_opened, result=result))
else:
html = ("<table>"
"<tr>"
"<th>Face State</th>"
"<th>Value</th>"
"</tr>"
"<tr>"
"<td>Result</td>"
"<td>{result}</td>"
"</tr>"
"</table>".format(result=result))
try:
image = Image.open(frame)
for face in r.json().get('faces'):
x1 = face.get('x1')
y1 = face.get('y1')
x2 = face.get('x2')
y2 = face.get('y2')
if x1 < 0:
x1 = 0
if y1 < 0:
y1 = 0
if x2 >= image.width:
x2 = image.width - 1
if y2 >= image.height:
y2 = image.height - 1
face_image = image.crop((x1, y1, x2, y2))
face_image_ratio = face_image.width / float(face_image.height)
resized_w = int(face_image_ratio * 150)
resized_h = 150
face_image = face_image.resize((int(resized_w), int(resized_h)))
if faces is None:
faces = face_image
else:
new_image = Image.new('RGB',(faces.width + face_image.width + 10, 150), (80,80,80))
new_image.paste(faces,(0,0))
new_image.paste(face_image,(faces.width + 10, 0))
faces = new_image.copy()
except:
pass
return [faces, html]
with gr.Blocks() as demo:
gr.Markdown(
"""
# KBY-AI
We offer SDKs for Face Recognition, Face Liveness Detection(Face Anti-Spoofing), and ID Card Recognition.<br/>
Besides that, we can provide several AI models and development services in machine learning.
## Simple Installation & Simple API
```
sudo docker pull kbyai/face-liveness-detection:latest
sudo docker run -e LICENSE="xxxxx" -p 8080:8080 -p 9000:9000 kbyai/face-liveness-detection:latest
```
## KYC Verification Demo
https://github.com/kby-ai/KYC-Verification
"""
)
with gr.TabItem("Face Liveness Detection"):
with gr.Row():
with gr.Column():
live_image_input = gr.Image(type='filepath')
gr.Examples(['live_examples/1.jpg', 'live_examples/2.jpg', 'live_examples/3.jpg', 'live_examples/4.jpg'],
inputs=live_image_input)
check_liveness_button = gr.Button("Check Liveness")
with gr.Column():
liveness_face_output = gr.Image(type="pil").style(height=150)
livness_result_output = gr.HTML()
check_liveness_button.click(check_liveness, inputs=live_image_input, outputs=[liveness_face_output, livness_result_output])
demo.launch(server_name="0.0.0.0", server_port=9000) |