-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathapp.py
199 lines (170 loc) · 8.04 KB
/
app.py
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
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
import gradio as gr
# from PIL import Image
# from modelscope_studio import encode_image, decode_image, call_demo_service
import json
import os
from modelscope.pipelines import pipeline
from modelscope.utils.constant import Tasks
import cv2
import face_preprocess
import imageio
###########################################
# 调试参数
###########################################
is_local_debug = False
debug_det_result_dir = './debug_det_result/'
debug_example_dir = './new_pic_wangzai/'
def save_det_result(img_path, src_img, raw_result):
from modelscope.utils.cv.image_utils import draw_face_detection_result
if not os.path.exists(debug_det_result_dir):
os.makedirs(debug_det_result_dir)
debug_src_img_path = debug_det_result_dir + 'det_src_' + os.path.split(img_path)[1]
# save src img
cv2.imwrite(debug_src_img_path, src_img)
# save det result img
dst_img = draw_face_detection_result(debug_src_img_path, raw_result)
cv2.imwrite(debug_det_result_dir + 'det_result_' + os.path.split(img_path)[1], dst_img)
###########################################
# 全局超参
###########################################
online_example_dir = '/images/'
online_example_id = ['id1/', 'id2/']
is_compress_input_img = True
# 视频参数
video_fps = 5
video_size = [224, 256]
face_template_size = [192, 224]
# 输入图像参数
compress_large_edge = 1280
# 融合渐变次数
blend_num = 3
# 原图停留次数
stop_num = 2
def compress_input_img(src_img):
dst_width = src_img.shape[1] if compress_large_edge > src_img.shape[1] else compress_large_edge
dst_height = int(src_img.shape[0] * dst_width / src_img.shape[1])
src_img = cv2.resize(src_img, (dst_width, dst_height), interpolation=cv2.INTER_AREA)
return src_img
###########################################
# gradio demo app 推断入口
###########################################
# gradio app demo 算法运行函数
def inference(input_files, input_param_gif_fps, input_param_gif_width, input_param_gif_height):
video_fps = int(input_param_gif_fps)
video_size[0] = int(input_param_gif_width)
video_size[1] = int(input_param_gif_height)
video_last_img = None
gif_img_lst = []
if input_files is None:
return None
input_files = [x.name for x in input_files]
print(input_files)
# 过程中间变量
valid_img_num = 0
# 绕过输入
if is_local_debug :
input_files = os.listdir(os.path.dirname(__file__) + debug_example_dir)
input_files.sort()
input_files = [os.path.dirname(__file__) + debug_example_dir + x for x in input_files]
for i in range(0, len(input_files)):
try:
print(input_files[i])
src_img_path = input_files[i]
src_img = cv2.imread(src_img_path)
if src_img is None:
continue
# # compress input image
if is_compress_input_img :
src_img = compress_input_img(src_img)
# get face detection and landmark result
facial_landmark_confidence_func = pipeline(Tasks.face_2d_keypoints, 'damo/cv_manual_facial-landmark-confidence_flcm')
raw_result = facial_landmark_confidence_func(src_img)
if raw_result is None:
continue
if float(raw_result['scores'][0]) < (1 - 0.145) :
print('landmark quality fail...')
continue
# import pdb;pdb.set_trace()
box_ldmk_str = ' '.join(str(x) for x in raw_result['boxes'][0]) \
+ ' ' + ' '.join(str(raw_result['keypoints'][0][x*2]) for x in range(5)) \
+ ' ' + ' '.join(str(raw_result['keypoints'][0][x*2+1]) for x in range(5))
# if you want to show the result, you can run
if is_local_debug :
save_det_result(src_img_path, src_img, raw_result)
# align face
warped_face = face_preprocess.align_face(src_img, box_ldmk_str, face_template_size, video_size)
if warped_face is None:
print('warp ' + src_img_path + ' error')
continue
# 获得渐变效果
if valid_img_num != 0:
last_face = video_last_img
cur_face = warped_face
for j in range(0, blend_num):
alpha = (j+1) / (blend_num+1)
beta = 1 - alpha
# face add weight
last_face = cv2.addWeighted(cur_face, alpha, last_face, beta, 0.)
gif_img_lst.append(cv2.cvtColor(last_face, cv2.COLOR_BGR2RGB))
# 获得停留效果
for j in range(0, stop_num):
gif_img_lst.append(cv2.cvtColor(warped_face, cv2.COLOR_BGR2RGB))
video_last_img = warped_face
valid_img_num += 1
except Exception:
print('process ' + input_files[i] + ' ' + Exception)
pass
if len(gif_img_lst) == 0:
return None
imageio.mimsave('output.gif', gif_img_lst, fps=video_fps)
return 'output.gif'
# gradio app 环境参数
css_style = "#fixed_size_img {height: 240px;} " \
"#overview {margin: auto;max-width: 600px; max-height: 400px;}"
title = "时光相册"
description = "<center> <p> 随意上传您心仪的照片合集（宝宝成长相册、同学毕业相册），通过人脸五官对齐，一键生成样貌变化效果图，点点鼠标即可下载分享给亲朋好友啦，立刻玩起来吧!!! </p> </center>"
###########################################
# gradio demo app
###########################################
with gr.Blocks(title=title, css=css_style) as demo:
gr.HTML('''
<div style="text-align: center; max-width: 720px; margin: 0 auto;">
<div
style="
display: inline-flex;
align-items: center;
gap: 0.8rem;
font-size: 1.75rem;
"
>
<h1 style="font-family: PingFangSC; font-weight: 500; line-height: 1.5em; font-size: 32px; margin-bottom: 7px;">
时光相册
</h1>
</div>
<img id="overview" alt="overview" src="https://mogface.oss-cn-zhangjiakou.aliyuncs.com/modelscope/face_album/template.gif" />
</div>
''')
gr.Markdown(description)
with gr.Row():
input_param_gif_fps = gr.Slider(1, 32, step =1, value =5, label="视频帧率（数值越大播放越快，非常有趣，试试看！！！）" )
input_param_gif_width = gr.Slider(224, 1280, step =8, value =224, label="视频宽度（数值越大横向背景越多）" )
input_param_gif_height = gr.Slider(256, 1280, step= 8, value=256,label="视频高度（数值越大纵向背景越多）")
with gr.Row():
input_files = gr.Files(elem_id="fixed_size_img", label="图片合集（上传同一人在不同时期的照片，暂不支持补、删图，修改合集建议刷新或清除后重传）")
img_output = gr.Image(type="filepath", elem_id="fixed_size_img", label="视频结果")
with gr.Row():
btn_submit = gr.Button(value="一键生成", elem_id="blue_btn")
# append all examples
examples = []
for i in range(0, len(online_example_id)):
example_id_dir = os.path.dirname(__file__) + online_example_dir + online_example_id[i]
example_id_names = os.listdir(example_id_dir)
example_id_names.sort()
example_id_names = [example_id_dir + x for x in example_id_names]
examples.append([example_id_names])
print(examples)
examples = gr.Examples(examples=examples, inputs=input_files, outputs=img_output, label="点击如下示例试玩", run_on_click=False)
btn_submit.click(inference, inputs=[input_files, input_param_gif_fps, input_param_gif_width, input_param_gif_height], outputs=img_output)
# btn_clear清除画布
if __name__ == "__main__":
demo.launch(share=False)