-
Notifications
You must be signed in to change notification settings - Fork 3
/
app.py
145 lines (121 loc) · 6.89 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
from argparse import ArgumentParser, BooleanOptionalAction
import gradio as gr
from PIL import Image, ImageOps
import cv2
import numpy as np
import matplotlib.pyplot as plt
from tools import depth, sam, spectralresidual, brightness, brightness_diff
from tools import depth
from utils import load_seg_model
import os
PROGRAM_NAME = 'AMeThyst'
path = os.getcwd()
output_dir = f"{path}/output"
input_dir = f"{path}/input"
model_dir = f"{path}/models/sam"
load_seg_model(model_dir)
def create_heatmap_overlay(image, result, alpha=0.5, save_path='heatmap_overlay.png'):
height, width = image.shape[:2]
image = np.array(image, dtype=np.uint8)
color_image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
plt.imshow(result, cmap='jet', alpha=alpha)
plt.axis('off')
plt.savefig(f'{path}/heatmap/heatmap.png', bbox_inches='tight', pad_inches=0)
plt.close()
heatmap = cv2.cvtColor(cv2.imread(f'{path}/heatmap/heatmap.png'), cv2.COLOR_BGR2RGB)
heatmap = cv2.resize(heatmap, (width, height))
cv2.imwrite(f'{path}/heatmap.png', heatmap)
overlay = cv2.addWeighted(color_image, 1-alpha, heatmap, alpha, 0)
cv2.imwrite(save_path, overlay)
return overlay
def plot_overlay(image, analysis_result):
#image = image.convert('RGB')
image_array = np.array(image)
image_array = cv2.cvtColor(image_array, cv2.COLOR_RGB2BGR)
image_overlay = create_heatmap_overlay(image_array, analysis_result, alpha=0.5, save_path='heatmap_overlay.png')
return image_overlay
def composite(results):
base = results[0][0] * results[0][1]
for result in results[1:]:
base = base + result[0] * result[1]
result = base/len(results)
return result
def analysis(
image,
sam_flag, depth_flag, spectralresidual_flag, brightness_flg, brightness_diff_flg,
sam_heatmap_weight, sam_window_size, points_per_side, min_mask_region_area, pred_iou_thresh, stability_score_thresh,
depth_heatmap_weight, depth_window_size,
spectralresidual_heatmap_weight,
brightness_heatmap_weight, brightness_window_size,
brightness_diff_heatmap_weight, brightness_diff_window_size
):
results = []
if sam_flag == True:
results.append((sam.process(model_dir, image, sam_window_size, points_per_side, pred_iou_thresh, stability_score_thresh, min_mask_region_area), sam_heatmap_weight))
if depth_flag == True:
results.append((depth.process(image, depth_window_size), depth_heatmap_weight))
if spectralresidual_flag == True:
results.append((spectralresidual.process(image), spectralresidual_heatmap_weight))
if brightness_flg == True:
results.append((brightness.process(image, brightness_window_size), brightness_heatmap_weight))
if brightness_diff_flg == True:
results.append((brightness_diff.process(image, brightness_diff_window_size), brightness_diff_heatmap_weight))
if len(results) > 1:
result = composite(results)
else:
result = results[0][0]
image_overlay = plot_overlay(image, result)
return image_overlay
def parse_args():
parser = ArgumentParser(prog=PROGRAM_NAME, description="Art Metrics Tools for hypothesis test.")
parser.add_argument('--inbrowser', action=BooleanOptionalAction, default=False, help="Gradio inbrowser")
parser.add_argument('--share', action=BooleanOptionalAction, default=True, help="Gradio Share")
parser.add_argument('--max_file_size', type=str, default=10 * gr.FileSize.MB, help="Gradio Upload MaxFileSize")
return parser.parse_args()
# Gradio UIの設定
enables = {}
sam_params = {}
with gr.Blocks(analytics_enabled=False, title=PROGRAM_NAME) as demo:
with gr.Row() as common:
with gr.Column() as sub:
input=gr.Image(type="pil")
submit=gr.Button("submit")
with gr.Accordion("Object Density") as sam_block:
sam_flag = gr.Checkbox(True, label="enable")
sam_heatmap_weight = gr.Slider(value = 1, minimum=0, maximum=1, step=0.05, label="heatmap_weight", randomize=True)
sam_window_size = gr.Number(value=64, label="window_size")
with gr.Accordion("SAM Parameter") as sam_param:
points_per_side = gr.Number(value=32, label="window_size")
min_mask_region_area=gr.Number(value=10, label="window_size")
pred_iou_thresh=gr.Slider(value = 0.8, minimum=0, maximum=1, step=0.05, label="pred_iou_thresh", randomize=True)
stability_score_thresh=gr.Slider(value = 0.8, minimum=0, maximum=1, step=0.05, label="stability_score_thresh", randomize=True)
with gr.Accordion("Subject Distance") as sam_block:
depth_flag = gr.Checkbox(False, label="enable")
depth_heatmap_weight = gr.Slider(value = 0, minimum=0, maximum=1, step=0.05, label="heatmap_weight", randomize=True)
depth_window_size = gr.Number(value=64, label="window_size")
with gr.Accordion("Spectral Residual") as sam_block:
spectralresidual_flag = gr.Checkbox(False, label="enable")
spectralresidual_heatmap_weight = gr.Slider(value = 0, minimum=0, maximum=1, step=0.05, label="heatmap_weight", randomize=True)
with gr.Accordion("Brightness") as sam_block:
brightness_flag = gr.Checkbox(False, label="enable")
brightness_heatmap_weight = gr.Slider(value = 0, minimum=0, maximum=1, step=0.05, label="heatmap_weight", randomize=True)
brightness_window_size = gr.Number(value=64, label="window_size")
with gr.Accordion("Brightness Diff") as sam_block:
brightness_diff_flag = gr.Checkbox(False, label="enable")
brightness_diff_heatmap_weight = gr.Slider(value = 0, minimum=0, maximum=1, step=0.05, label="heatmap_weight", randomize=True)
brightness_diff_window_size = gr.Number(value=64, label="window_size")
output=gr.Image(type="pil")
submit.click(
fn=analysis, inputs=[
input,
sam_flag, depth_flag, spectralresidual_flag, brightness_flag, brightness_diff_flag,
sam_heatmap_weight, sam_window_size, points_per_side, min_mask_region_area, pred_iou_thresh, stability_score_thresh,
depth_heatmap_weight, depth_window_size,
spectralresidual_heatmap_weight,
brightness_heatmap_weight, brightness_window_size,
brightness_diff_heatmap_weight, brightness_diff_window_size
], outputs=output)
# コマンドライン引数の解析
args = parse_args()
# インターフェースの起動
demo.launch(inbrowser=args.inbrowser, share=args.share, max_file_size=args.max_file_size)