Spaces:
Running
Running
Commit
·
ea07ffb
1
Parent(s):
b46ef65
app
Browse files
app.py
CHANGED
@@ -1,7 +1,696 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
import gradio as gr
|
2 |
|
3 |
-
|
4 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
5 |
|
6 |
-
demo
|
7 |
-
demo.launch()
|
|
|
1 |
+
import argparse
|
2 |
+
import os
|
3 |
+
import random
|
4 |
+
from collections import defaultdict
|
5 |
+
|
6 |
+
import cv2
|
7 |
+
import re
|
8 |
+
|
9 |
+
import numpy as np
|
10 |
+
from PIL import Image
|
11 |
+
import torch
|
12 |
+
import html
|
13 |
import gradio as gr
|
14 |
|
15 |
+
import torchvision.transforms as T
|
16 |
+
import torch.backends.cudnn as cudnn
|
17 |
+
|
18 |
+
from minigpt4.common.config import Config
|
19 |
+
|
20 |
+
from minigpt4.common.registry import registry
|
21 |
+
from minigpt4.conversation.conversation import Conversation, SeparatorStyle, Chat
|
22 |
+
|
23 |
+
# imports modules for registration
|
24 |
+
from minigpt4.datasets.builders import *
|
25 |
+
from minigpt4.models import *
|
26 |
+
from minigpt4.processors import *
|
27 |
+
from minigpt4.runners import *
|
28 |
+
from minigpt4.tasks import *
|
29 |
+
|
30 |
+
|
31 |
+
def parse_args():
|
32 |
+
parser = argparse.ArgumentParser(description="Demo")
|
33 |
+
parser.add_argument("--cfg-path", default='eval_configs/demo.yaml',
|
34 |
+
help="path to configuration file.")
|
35 |
+
parser.add_argument(
|
36 |
+
"--options",
|
37 |
+
nargs="+",
|
38 |
+
help="override some settings in the used config, the key-value pair "
|
39 |
+
"in xxx=yyy format will be merged into config file (deprecate), "
|
40 |
+
"change to --cfg-options instead.",
|
41 |
+
)
|
42 |
+
args = parser.parse_args()
|
43 |
+
return args
|
44 |
+
|
45 |
+
|
46 |
+
random.seed(42)
|
47 |
+
np.random.seed(42)
|
48 |
+
torch.manual_seed(42)
|
49 |
+
|
50 |
+
cudnn.benchmark = False
|
51 |
+
cudnn.deterministic = True
|
52 |
+
|
53 |
+
print('Initializing Chat')
|
54 |
+
args = parse_args()
|
55 |
+
cfg = Config(args)
|
56 |
+
|
57 |
+
device = 'cuda'
|
58 |
+
|
59 |
+
model_config = cfg.model_cfg
|
60 |
+
|
61 |
+
print("model_config:", model_config)
|
62 |
+
model_cls = registry.get_model_class(model_config.arch)
|
63 |
+
model = model_cls.from_config(model_config).to(device)
|
64 |
+
bounding_box_size = 100
|
65 |
+
|
66 |
+
vis_processor_cfg = cfg.datasets_cfg.feature_face_caption.vis_processor.train
|
67 |
+
vis_processor = registry.get_processor_class(vis_processor_cfg.name).from_config(vis_processor_cfg)
|
68 |
+
|
69 |
+
model = model.eval()
|
70 |
+
|
71 |
+
CONV_VISION = Conversation(
|
72 |
+
system="",
|
73 |
+
roles=(r"<s>[INST] ", r" [/INST]"),
|
74 |
+
messages=[],
|
75 |
+
offset=2,
|
76 |
+
sep_style=SeparatorStyle.SINGLE,
|
77 |
+
sep="",
|
78 |
+
)
|
79 |
+
|
80 |
+
|
81 |
+
def extract_substrings(string):
|
82 |
+
# first check if there is no-finished bracket
|
83 |
+
index = string.rfind('}')
|
84 |
+
if index != -1:
|
85 |
+
string = string[:index + 1]
|
86 |
+
|
87 |
+
pattern = r'<p>(.*?)\}(?!<)'
|
88 |
+
matches = re.findall(pattern, string)
|
89 |
+
substrings = [match for match in matches]
|
90 |
+
|
91 |
+
return substrings
|
92 |
+
|
93 |
+
|
94 |
+
def is_overlapping(rect1, rect2):
|
95 |
+
x1, y1, x2, y2 = rect1
|
96 |
+
x3, y3, x4, y4 = rect2
|
97 |
+
return not (x2 < x3 or x1 > x4 or y2 < y3 or y1 > y4)
|
98 |
+
|
99 |
+
|
100 |
+
def computeIoU(bbox1, bbox2):
|
101 |
+
x1, y1, x2, y2 = bbox1
|
102 |
+
x3, y3, x4, y4 = bbox2
|
103 |
+
intersection_x1 = max(x1, x3)
|
104 |
+
intersection_y1 = max(y1, y3)
|
105 |
+
intersection_x2 = min(x2, x4)
|
106 |
+
intersection_y2 = min(y2, y4)
|
107 |
+
intersection_area = max(0, intersection_x2 - intersection_x1 + 1) * max(0, intersection_y2 - intersection_y1 + 1)
|
108 |
+
bbox1_area = (x2 - x1 + 1) * (y2 - y1 + 1)
|
109 |
+
bbox2_area = (x4 - x3 + 1) * (y4 - y3 + 1)
|
110 |
+
union_area = bbox1_area + bbox2_area - intersection_area
|
111 |
+
iou = intersection_area / union_area
|
112 |
+
return iou
|
113 |
+
|
114 |
+
|
115 |
+
def save_tmp_img(visual_img):
|
116 |
+
file_name = "".join([str(random.randint(0, 9)) for _ in range(5)]) + ".jpg"
|
117 |
+
file_path = "/tmp/gradio" + file_name
|
118 |
+
visual_img.save(file_path)
|
119 |
+
return file_path
|
120 |
+
|
121 |
+
|
122 |
+
def mask2bbox(mask):
|
123 |
+
if mask is None:
|
124 |
+
return ''
|
125 |
+
mask = mask.resize([100, 100], resample=Image.NEAREST)
|
126 |
+
mask = np.array(mask)[:, :, 0]
|
127 |
+
|
128 |
+
rows = np.any(mask, axis=1)
|
129 |
+
cols = np.any(mask, axis=0)
|
130 |
+
|
131 |
+
if rows.sum():
|
132 |
+
# Get the top, bottom, left, and right boundaries
|
133 |
+
rmin, rmax = np.where(rows)[0][[0, -1]]
|
134 |
+
cmin, cmax = np.where(cols)[0][[0, -1]]
|
135 |
+
bbox = '{{<{}><{}><{}><{}>}}'.format(cmin, rmin, cmax, rmax)
|
136 |
+
else:
|
137 |
+
bbox = ''
|
138 |
+
|
139 |
+
return bbox
|
140 |
+
|
141 |
+
|
142 |
+
def escape_markdown(text):
|
143 |
+
# List of Markdown special characters that need to be escaped
|
144 |
+
md_chars = ['<', '>']
|
145 |
+
|
146 |
+
# Escape each special character
|
147 |
+
for char in md_chars:
|
148 |
+
text = text.replace(char, '\\' + char)
|
149 |
+
|
150 |
+
return text
|
151 |
+
|
152 |
+
|
153 |
+
def reverse_escape(text):
|
154 |
+
md_chars = ['\\<', '\\>']
|
155 |
+
|
156 |
+
for char in md_chars:
|
157 |
+
text = text.replace(char, char[1:])
|
158 |
+
|
159 |
+
return text
|
160 |
+
|
161 |
+
|
162 |
+
colors = [
|
163 |
+
(255, 0, 0),
|
164 |
+
(0, 255, 0),
|
165 |
+
(0, 0, 255),
|
166 |
+
(210, 210, 0),
|
167 |
+
(255, 0, 255),
|
168 |
+
(0, 255, 255),
|
169 |
+
(114, 128, 250),
|
170 |
+
(0, 165, 255),
|
171 |
+
(0, 128, 0),
|
172 |
+
(144, 238, 144),
|
173 |
+
(238, 238, 175),
|
174 |
+
(255, 191, 0),
|
175 |
+
(0, 128, 0),
|
176 |
+
(226, 43, 138),
|
177 |
+
(255, 0, 255),
|
178 |
+
(0, 215, 255),
|
179 |
+
]
|
180 |
+
|
181 |
+
color_map = {
|
182 |
+
f"{color_id}": f"#{hex(color[2])[2:].zfill(2)}{hex(color[1])[2:].zfill(2)}{hex(color[0])[2:].zfill(2)}" for
|
183 |
+
color_id, color in enumerate(colors)
|
184 |
+
}
|
185 |
+
|
186 |
+
used_colors = colors
|
187 |
+
|
188 |
+
def get_first_frame(video_path):
|
189 |
+
cap = cv2.VideoCapture(video_path)
|
190 |
+
|
191 |
+
if not cap.isOpened():
|
192 |
+
print("Error: Cannot open video.")
|
193 |
+
return None
|
194 |
+
|
195 |
+
ret, frame = cap.read()
|
196 |
+
cap.release()
|
197 |
+
|
198 |
+
if ret:
|
199 |
+
return frame
|
200 |
+
else:
|
201 |
+
print("Error: Cannot read frame from video.")
|
202 |
+
return None
|
203 |
+
|
204 |
+
def visualize_all_bbox_together(image, generation):
|
205 |
+
if image is None:
|
206 |
+
return None, ''
|
207 |
+
|
208 |
+
if isinstance(image, str): # is a image path
|
209 |
+
raw_image = get_first_frame(image)
|
210 |
+
frame_rgb = cv2.cvtColor(raw_image, cv2.COLOR_BGR2RGB)
|
211 |
+
image = Image.fromarray(frame_rgb)
|
212 |
+
|
213 |
+
generation = html.unescape(generation)
|
214 |
+
|
215 |
+
image_width, image_height = image.size
|
216 |
+
image = image.resize([500, int(500 / image_width * image_height)])
|
217 |
+
image_width, image_height = image.size
|
218 |
+
|
219 |
+
string_list = extract_substrings(generation)
|
220 |
+
if string_list: # it is grounding or detection
|
221 |
+
mode = 'all'
|
222 |
+
entities = defaultdict(list)
|
223 |
+
i = 0
|
224 |
+
j = 0
|
225 |
+
for string in string_list:
|
226 |
+
try:
|
227 |
+
obj, string = string.split('</p>')
|
228 |
+
except ValueError:
|
229 |
+
print('wrong string: ', string)
|
230 |
+
continue
|
231 |
+
bbox_list = string.split('<delim>')
|
232 |
+
flag = False
|
233 |
+
for bbox_string in bbox_list:
|
234 |
+
integers = re.findall(r'-?\d+', bbox_string)
|
235 |
+
if len(integers) == 4:
|
236 |
+
x0, y0, x1, y1 = int(integers[0]), int(integers[1]), int(integers[2]), int(integers[3])
|
237 |
+
left = x0 / bounding_box_size * image_width
|
238 |
+
bottom = y0 / bounding_box_size * image_height
|
239 |
+
right = x1 / bounding_box_size * image_width
|
240 |
+
top = y1 / bounding_box_size * image_height
|
241 |
+
|
242 |
+
entities[obj].append([left, bottom, right, top])
|
243 |
+
|
244 |
+
j += 1
|
245 |
+
flag = True
|
246 |
+
if flag:
|
247 |
+
i += 1
|
248 |
+
else:
|
249 |
+
integers = re.findall(r'-?\d+', generation)
|
250 |
+
|
251 |
+
if len(integers) == 4: # it is refer
|
252 |
+
mode = 'single'
|
253 |
+
|
254 |
+
entities = list()
|
255 |
+
x0, y0, x1, y1 = int(integers[0]), int(integers[1]), int(integers[2]), int(integers[3])
|
256 |
+
left = x0 / bounding_box_size * image_width
|
257 |
+
bottom = y0 / bounding_box_size * image_height
|
258 |
+
right = x1 / bounding_box_size * image_width
|
259 |
+
top = y1 / bounding_box_size * image_height
|
260 |
+
entities.append([left, bottom, right, top])
|
261 |
+
else:
|
262 |
+
# don't detect any valid bbox to visualize
|
263 |
+
return None, ''
|
264 |
+
|
265 |
+
if len(entities) == 0:
|
266 |
+
return None, ''
|
267 |
+
|
268 |
+
if isinstance(image, Image.Image):
|
269 |
+
image_h = image.height
|
270 |
+
image_w = image.width
|
271 |
+
image = np.array(image)
|
272 |
+
|
273 |
+
elif isinstance(image, str):
|
274 |
+
if os.path.exists(image):
|
275 |
+
pil_img = Image.open(image).convert("RGB")
|
276 |
+
image = np.array(pil_img)[:, :, [2, 1, 0]]
|
277 |
+
image_h = pil_img.height
|
278 |
+
image_w = pil_img.width
|
279 |
+
else:
|
280 |
+
raise ValueError(f"invaild image path, {image}")
|
281 |
+
elif isinstance(image, torch.Tensor):
|
282 |
+
|
283 |
+
image_tensor = image.cpu()
|
284 |
+
reverse_norm_mean = torch.tensor([0.48145466, 0.4578275, 0.40821073])[:, None, None]
|
285 |
+
reverse_norm_std = torch.tensor([0.26862954, 0.26130258, 0.27577711])[:, None, None]
|
286 |
+
image_tensor = image_tensor * reverse_norm_std + reverse_norm_mean
|
287 |
+
pil_img = T.ToPILImage()(image_tensor)
|
288 |
+
image_h = pil_img.height
|
289 |
+
image_w = pil_img.width
|
290 |
+
image = np.array(pil_img)[:, :, [2, 1, 0]]
|
291 |
+
else:
|
292 |
+
raise ValueError(f"invaild image format, {type(image)} for {image}")
|
293 |
+
|
294 |
+
indices = list(range(len(entities)))
|
295 |
+
|
296 |
+
new_image = image.copy()
|
297 |
+
|
298 |
+
previous_bboxes = []
|
299 |
+
# size of text
|
300 |
+
text_size = 0.5
|
301 |
+
# thickness of text
|
302 |
+
text_line = 1 # int(max(1 * min(image_h, image_w) / 512, 1))
|
303 |
+
box_line = 2
|
304 |
+
(c_width, text_height), _ = cv2.getTextSize("F", cv2.FONT_HERSHEY_COMPLEX, text_size, text_line)
|
305 |
+
base_height = int(text_height * 0.675)
|
306 |
+
text_offset_original = text_height - base_height
|
307 |
+
text_spaces = 2
|
308 |
+
|
309 |
+
# num_bboxes = sum(len(x[-1]) for x in entities)
|
310 |
+
used_colors = colors # random.sample(colors, k=num_bboxes)
|
311 |
+
|
312 |
+
color_id = -1
|
313 |
+
for entity_idx, entity_name in enumerate(entities):
|
314 |
+
if mode == 'single' or mode == 'identify':
|
315 |
+
bboxes = entity_name
|
316 |
+
bboxes = [bboxes]
|
317 |
+
else:
|
318 |
+
bboxes = entities[entity_name]
|
319 |
+
color_id += 1
|
320 |
+
for bbox_id, (x1_norm, y1_norm, x2_norm, y2_norm) in enumerate(bboxes):
|
321 |
+
skip_flag = False
|
322 |
+
orig_x1, orig_y1, orig_x2, orig_y2 = int(x1_norm), int(y1_norm), int(x2_norm), int(y2_norm)
|
323 |
+
|
324 |
+
color = used_colors[entity_idx % len(used_colors)] # tuple(np.random.randint(0, 255, size=3).tolist())
|
325 |
+
new_image = cv2.rectangle(new_image, (orig_x1, orig_y1), (orig_x2, orig_y2), color, box_line)
|
326 |
+
|
327 |
+
if mode == 'all':
|
328 |
+
l_o, r_o = box_line // 2 + box_line % 2, box_line // 2 + box_line % 2 + 1
|
329 |
+
|
330 |
+
x1 = orig_x1 - l_o
|
331 |
+
y1 = orig_y1 - l_o
|
332 |
+
|
333 |
+
if y1 < text_height + text_offset_original + 2 * text_spaces:
|
334 |
+
y1 = orig_y1 + r_o + text_height + text_offset_original + 2 * text_spaces
|
335 |
+
x1 = orig_x1 + r_o
|
336 |
+
|
337 |
+
# add text background
|
338 |
+
(text_width, text_height), _ = cv2.getTextSize(f" {entity_name}", cv2.FONT_HERSHEY_COMPLEX, text_size,
|
339 |
+
text_line)
|
340 |
+
text_bg_x1, text_bg_y1, text_bg_x2, text_bg_y2 = x1, y1 - (
|
341 |
+
text_height + text_offset_original + 2 * text_spaces), x1 + text_width, y1
|
342 |
+
|
343 |
+
for prev_bbox in previous_bboxes:
|
344 |
+
if computeIoU((text_bg_x1, text_bg_y1, text_bg_x2, text_bg_y2), prev_bbox['bbox']) > 0.95 and \
|
345 |
+
prev_bbox['phrase'] == entity_name:
|
346 |
+
skip_flag = True
|
347 |
+
break
|
348 |
+
while is_overlapping((text_bg_x1, text_bg_y1, text_bg_x2, text_bg_y2), prev_bbox['bbox']):
|
349 |
+
text_bg_y1 += (text_height + text_offset_original + 2 * text_spaces)
|
350 |
+
text_bg_y2 += (text_height + text_offset_original + 2 * text_spaces)
|
351 |
+
y1 += (text_height + text_offset_original + 2 * text_spaces)
|
352 |
+
|
353 |
+
if text_bg_y2 >= image_h:
|
354 |
+
text_bg_y1 = max(0, image_h - (text_height + text_offset_original + 2 * text_spaces))
|
355 |
+
text_bg_y2 = image_h
|
356 |
+
y1 = image_h
|
357 |
+
break
|
358 |
+
if not skip_flag:
|
359 |
+
alpha = 0.5
|
360 |
+
for i in range(text_bg_y1, text_bg_y2):
|
361 |
+
for j in range(text_bg_x1, text_bg_x2):
|
362 |
+
if i < image_h and j < image_w:
|
363 |
+
if j < text_bg_x1 + 1.35 * c_width:
|
364 |
+
# original color
|
365 |
+
bg_color = color
|
366 |
+
else:
|
367 |
+
# white
|
368 |
+
bg_color = [255, 255, 255]
|
369 |
+
new_image[i, j] = (alpha * new_image[i, j] + (1 - alpha) * np.array(bg_color)).astype(
|
370 |
+
np.uint8)
|
371 |
+
|
372 |
+
cv2.putText(
|
373 |
+
new_image, f" {entity_name}", (x1, y1 - text_offset_original - 1 * text_spaces),
|
374 |
+
cv2.FONT_HERSHEY_COMPLEX, text_size, (0, 0, 0), text_line, cv2.LINE_AA
|
375 |
+
)
|
376 |
+
|
377 |
+
previous_bboxes.append(
|
378 |
+
{'bbox': (text_bg_x1, text_bg_y1, text_bg_x2, text_bg_y2), 'phrase': entity_name})
|
379 |
+
|
380 |
+
if mode == 'all':
|
381 |
+
def color_iterator(colors):
|
382 |
+
while True:
|
383 |
+
for color in colors:
|
384 |
+
yield color
|
385 |
+
|
386 |
+
color_gen = color_iterator(colors)
|
387 |
+
|
388 |
+
# Add colors to phrases and remove <p></p>
|
389 |
+
def colored_phrases(match):
|
390 |
+
phrase = match.group(1)
|
391 |
+
color = next(color_gen)
|
392 |
+
return f'<span style="color:rgb{color}">{phrase}</span>'
|
393 |
+
|
394 |
+
generation = re.sub(r'{<\d+><\d+><\d+><\d+>}|<delim>', '', generation)
|
395 |
+
generation_colored = re.sub(r'<p>(.*?)</p>', colored_phrases, generation)
|
396 |
+
else:
|
397 |
+
generation_colored = ''
|
398 |
+
|
399 |
+
pil_image = Image.fromarray(new_image)
|
400 |
+
return pil_image, generation_colored
|
401 |
+
|
402 |
+
|
403 |
+
def gradio_reset(chat_state, img_list):
|
404 |
+
if chat_state is not None:
|
405 |
+
chat_state.messages = []
|
406 |
+
if img_list is not None:
|
407 |
+
img_list = []
|
408 |
+
return None, gr.update(value=None, interactive=True), gr.update(placeholder='Upload your image and chat',
|
409 |
+
interactive=True), chat_state, img_list
|
410 |
+
|
411 |
+
|
412 |
+
def image_upload_trigger(upload_flag, replace_flag, img_list):
|
413 |
+
# set the upload flag to true when receive a new image.
|
414 |
+
# if there is an old image (and old conversation), set the replace flag to true to reset the conv later.
|
415 |
+
upload_flag = 1
|
416 |
+
if img_list:
|
417 |
+
replace_flag = 1
|
418 |
+
return upload_flag, replace_flag
|
419 |
+
|
420 |
+
|
421 |
+
def example_trigger(text_input, image, upload_flag, replace_flag, img_list):
|
422 |
+
# set the upload flag to true when receive a new image.
|
423 |
+
# if there is an old image (and old conversation), set the replace flag to true to reset the conv later.
|
424 |
+
upload_flag = 1
|
425 |
+
if img_list or replace_flag == 1:
|
426 |
+
replace_flag = 1
|
427 |
+
|
428 |
+
return upload_flag, replace_flag
|
429 |
+
|
430 |
+
|
431 |
+
def gradio_ask(user_message, chatbot, chat_state, gr_img, img_list, upload_flag, replace_flag):
|
432 |
+
print("+++gradio_ask+++")
|
433 |
+
|
434 |
+
if len(user_message) == 0:
|
435 |
+
text_box_show = 'Input should not be empty!'
|
436 |
+
else:
|
437 |
+
text_box_show = ''
|
438 |
+
|
439 |
+
print('user_message:', user_message)
|
440 |
+
print('chatbot:', chatbot)
|
441 |
+
print('chat_state:', chat_state)
|
442 |
+
|
443 |
+
|
444 |
+
if isinstance(gr_img, dict):
|
445 |
+
gr_img, mask = gr_img['image'], gr_img['mask']
|
446 |
+
else:
|
447 |
+
mask = None
|
448 |
+
|
449 |
+
if '[identify]' in user_message:
|
450 |
+
# check if user provide bbox in the text input
|
451 |
+
integers = re.findall(r'-?\d+', user_message)
|
452 |
+
if len(integers) != 4: # no bbox in text
|
453 |
+
bbox = mask2bbox(mask)
|
454 |
+
user_message = user_message + bbox
|
455 |
+
|
456 |
+
if chat_state is None:
|
457 |
+
chat_state = CONV_VISION.copy()
|
458 |
+
|
459 |
+
if upload_flag:
|
460 |
+
if replace_flag:
|
461 |
+
chat_state = CONV_VISION.copy() # new image, reset everything
|
462 |
+
replace_flag = 0
|
463 |
+
chatbot = []
|
464 |
+
img_list = []
|
465 |
+
llm_message = chat.upload_img(gr_img, chat_state, img_list)
|
466 |
+
upload_flag = 0
|
467 |
+
|
468 |
+
chat.ask(user_message, chat_state)
|
469 |
+
print('user_message: ', user_message)
|
470 |
+
print('chat_state: ', chat_state)
|
471 |
+
|
472 |
+
chatbot = chatbot + [[user_message, None]]
|
473 |
+
|
474 |
+
if '[identify]' in user_message:
|
475 |
+
visual_img, _ = visualize_all_bbox_together(gr_img, user_message)
|
476 |
+
if visual_img is not None:
|
477 |
+
file_path = save_tmp_img(visual_img)
|
478 |
+
chatbot = chatbot + [[(file_path,), None]]
|
479 |
+
|
480 |
+
return text_box_show, chatbot, chat_state, img_list, upload_flag, replace_flag
|
481 |
+
|
482 |
+
|
483 |
+
def gradio_answer(chatbot, chat_state, img_list, temperature):
|
484 |
+
print("--gradio_answer--")
|
485 |
+
# print('img_list: ', img_list)
|
486 |
+
llm_message = chat.answer(conv=chat_state,
|
487 |
+
img_list=img_list,
|
488 |
+
temperature=temperature,
|
489 |
+
max_new_tokens=500,
|
490 |
+
max_length=2000)[0]
|
491 |
+
chatbot[-1][1] = llm_message
|
492 |
+
print('gradio_answer: ', llm_message)
|
493 |
+
|
494 |
+
return chatbot, chat_state
|
495 |
+
|
496 |
+
def process_english_text(text):
|
497 |
+
if len(text) < 2:
|
498 |
+
return text
|
499 |
+
text = text[0].upper() + text[1:]
|
500 |
+
|
501 |
+
sentences = text.split('. ')
|
502 |
+
corrected_sentences = [s.capitalize() for s in sentences]
|
503 |
+
text = '. '.join(corrected_sentences)
|
504 |
+
|
505 |
+
if text.endswith(','):
|
506 |
+
text = text[:-1]
|
507 |
+
if not text.endswith('.'):
|
508 |
+
text += '.'
|
509 |
+
|
510 |
+
return text
|
511 |
+
|
512 |
+
|
513 |
+
def gradio_stream_answer(chatbot, chat_state, img_list, temperature):
|
514 |
+
print('---gradio_stream_answer---')
|
515 |
+
if len(img_list) > 0:
|
516 |
+
if not isinstance(img_list[0], torch.Tensor):
|
517 |
+
chat.encode_img(img_list)
|
518 |
+
print(chat)
|
519 |
+
streamer = chat.stream_answer(conv=chat_state,
|
520 |
+
img_list=img_list,
|
521 |
+
temperature=temperature,
|
522 |
+
max_new_tokens=500,
|
523 |
+
max_length=2000)
|
524 |
+
output = ''
|
525 |
+
print('streamer:', streamer)
|
526 |
+
for new_output in streamer:
|
527 |
+
escapped = escape_markdown(new_output)
|
528 |
+
output += escapped
|
529 |
+
chatbot[-1][1] = output
|
530 |
+
chatbot[-1][1] = process_english_text(chatbot[-1][1])
|
531 |
+
yield chatbot, chat_state
|
532 |
+
chat_state.messages[-1][1] = '</s>'
|
533 |
+
print('output:', output)
|
534 |
+
return chatbot, chat_state
|
535 |
+
|
536 |
+
|
537 |
+
def gradio_visualize(chatbot, gr_img):
|
538 |
+
if isinstance(gr_img, dict):
|
539 |
+
gr_img, mask = gr_img['image'], gr_img['mask']
|
540 |
+
|
541 |
+
unescaped = reverse_escape(chatbot[-1][1])
|
542 |
+
visual_img, generation_color = visualize_all_bbox_together(gr_img, unescaped)
|
543 |
+
if visual_img is not None:
|
544 |
+
if len(generation_color):
|
545 |
+
chatbot[-1][1] = generation_color
|
546 |
+
file_path = save_tmp_img(visual_img)
|
547 |
+
chatbot = chatbot + [[None, (file_path,)]]
|
548 |
+
|
549 |
+
return chatbot
|
550 |
+
|
551 |
+
|
552 |
+
def gradio_taskselect(idx):
|
553 |
+
prompt_list = [
|
554 |
+
'',
|
555 |
+
'[reason] ',
|
556 |
+
'[emotion] ',
|
557 |
+
'[visual] ',
|
558 |
+
'[audio] '
|
559 |
+
]
|
560 |
+
instruct_list = [
|
561 |
+
'**Hint:** Type in whatever you want',
|
562 |
+
'**Hint:** Send the command to multimodal emotion reasoning',
|
563 |
+
'**Hint:** Send the command to multimodal emotion recognition',
|
564 |
+
'**Hint:** Send the command to generate visual description',
|
565 |
+
'**Hint:** Send the command to generate audio description'
|
566 |
+
]
|
567 |
+
return prompt_list[idx], instruct_list[idx]
|
568 |
+
|
569 |
+
|
570 |
+
|
571 |
+
|
572 |
+
chat = Chat(model, vis_processor, device=device)
|
573 |
+
|
574 |
+
title = """<h1 align="center">Emotion-LLaMA Demo</h1>"""
|
575 |
+
description = 'Welcome to Our Emotion-LLaMA Chatbot Demo!'
|
576 |
+
article = """<p><a href='https://anonymous.4open.science/r/Emotion-LLaMA'><img src='https://img.shields.io/badge/Project-Page-Green'></a></p>"""
|
577 |
+
|
578 |
+
introduction = '''
|
579 |
+
For Abilities Involging Multimodal Emotion Understanding:
|
580 |
+
1. Reason: Click **Send** to generate a multimodal emotion description.
|
581 |
+
2. Emotion: Click **Send** to generate an emotion label.
|
582 |
+
3. Visual: Click **Send** to generate a visual description.
|
583 |
+
4. Audio: Click **Send** to generate an audio description.
|
584 |
+
5. No Tag: Input whatever you want and click **Send** without any tagging.
|
585 |
+
|
586 |
+
You can also simply chat in free form!
|
587 |
+
'''
|
588 |
+
|
589 |
+
text_input = gr.Textbox(placeholder='Upload your image and chat', interactive=True, show_label=False, container=False, scale=8)
|
590 |
+
with gr.Blocks() as demo:
|
591 |
+
gr.Markdown(title)
|
592 |
+
# gr.Markdown(description)
|
593 |
+
gr.Markdown(article)
|
594 |
+
|
595 |
+
with gr.Row():
|
596 |
+
with gr.Column(scale=0.5):
|
597 |
+
# image = gr.Image(type="pil", tool='sketch', brush_radius=20)
|
598 |
+
image = gr.Video(sources=["upload", "webcam"])
|
599 |
+
|
600 |
+
temperature = gr.Slider(
|
601 |
+
minimum=0.1,
|
602 |
+
maximum=1.5,
|
603 |
+
value=0.2,
|
604 |
+
step=0.1,
|
605 |
+
interactive=True,
|
606 |
+
label="Temperature",
|
607 |
+
)
|
608 |
+
|
609 |
+
clear = gr.Button("Restart")
|
610 |
+
|
611 |
+
gr.Markdown(introduction)
|
612 |
+
|
613 |
+
with gr.Column():
|
614 |
+
chat_state = gr.State(value=None)
|
615 |
+
img_list = gr.State(value=[])
|
616 |
+
chatbot = gr.Chatbot(label='Emotion-LLaMA')
|
617 |
+
|
618 |
+
dataset = gr.Dataset(
|
619 |
+
components=[gr.Textbox(visible=False)],
|
620 |
+
samples=[['No Tag'], ['reason'], ['emotion'], ['visual'], ['audio']],
|
621 |
+
type="index",
|
622 |
+
label='Task Shortcuts',
|
623 |
+
)
|
624 |
+
task_inst = gr.Markdown('**Hint:** Upload your video and chat')
|
625 |
+
with gr.Row():
|
626 |
+
text_input.render()
|
627 |
+
send = gr.Button("Send", variant='primary', size='sm', scale=1)
|
628 |
+
|
629 |
+
upload_flag = gr.State(value=0)
|
630 |
+
replace_flag = gr.State(value=0)
|
631 |
+
image.upload(image_upload_trigger, [upload_flag, replace_flag, img_list], [upload_flag, replace_flag])
|
632 |
+
|
633 |
+
with gr.Row():
|
634 |
+
with gr.Column():
|
635 |
+
gr.Examples(examples=[
|
636 |
+
["examples/samplenew_00004251.mp4", "[detection] face", upload_flag, replace_flag, img_list],
|
637 |
+
["examples/sample_00000338.mp4", "The person in video says: Oh no, my phone and wallet are all in my bag. [emotion] Please determine which emotion label in the video represents: happy, sad, neutral, angry, worried, surprise.", upload_flag, replace_flag, img_list],
|
638 |
+
["examples/sample_00000669.mp4", "The person in video says: Why are you looking at me like this? It's just a woman, so you have to have something to do with me. [emotion] Determine the emotional state shown in the video, choosing from happy, sad, neutral, angry, worried, or surprise.", upload_flag, replace_flag, img_list],
|
639 |
+
["examples/sample_00003462.mp4", "The person in video says: Do you believe that you push me around? [emotion] Assess and label the emotion evident in the video: could it be happy, sad, neutral, angry, worried, surprise?", upload_flag, replace_flag, img_list],
|
640 |
+
["examples/sample_00000727.mp4", "The person in video says: No, this, I have to get up! You, I'm sorry, everyone. I'm sorry, it's from the German side. [emotion] Identify the displayed emotion in the video: is it happy, sad, neutral, angry, worried, or surprise?", upload_flag, replace_flag, img_list],
|
641 |
+
["examples/samplenew_00061200.mp4", "The person in video says: Me: I'm not going in anymore, scared. [emotion] Identify the displayed emotion in the video: is it happy, sad, neutral, angry, fear, contempt, doubt, worried, or surprise?", upload_flag, replace_flag, img_list],
|
642 |
+
], inputs=[image, text_input, upload_flag, replace_flag, img_list], fn=example_trigger,
|
643 |
+
outputs=[upload_flag, replace_flag])
|
644 |
+
with gr.Column():
|
645 |
+
gr.Examples(examples=[
|
646 |
+
["examples/samplenew_00051251.mp4", "In what state is the person in the video, say the following: \"Do you really think so?\"", upload_flag, replace_flag, img_list],
|
647 |
+
["examples/sample_00004735.mp4", "[visual] What are the emotions of the woman in the video?", upload_flag, replace_flag, img_list],
|
648 |
+
["examples/sample_00002422.mp4", "[audio] Analyze the speaker's voice in the video.", upload_flag, replace_flag, img_list],
|
649 |
+
["examples/sample_00001073.mp4", "The person in video says: Make him different from before. I like the way you are now. [reason] Please analyze all the clues in the video and reason out the emotional label of the person in the video.", upload_flag, replace_flag, img_list],
|
650 |
+
["examples/sample_00004671.mp4", "The person in video says: Won't you? Impossible! Fan Xiaomei is not such a person. [reason] What are the facial expressions and vocal tone used in the video? What is the intended meaning behind his words? Which emotion does this reflect?", upload_flag, replace_flag, img_list],
|
651 |
+
["examples/sample_00005854.mp4", "The person in video says: Bastard! Boss, you don't choose, you prefer. [reason] Please integrate information from various modalities to infer the emotional category of the person in the video.", upload_flag, replace_flag, img_list],
|
652 |
+
], inputs=[image, text_input, upload_flag, replace_flag, img_list], fn=example_trigger,
|
653 |
+
outputs=[upload_flag, replace_flag])
|
654 |
+
|
655 |
+
dataset.click(
|
656 |
+
gradio_taskselect,
|
657 |
+
inputs=[dataset],
|
658 |
+
outputs=[text_input, task_inst],
|
659 |
+
show_progress="hidden",
|
660 |
+
postprocess=False,
|
661 |
+
queue=False,
|
662 |
+
)
|
663 |
+
|
664 |
+
text_input.submit(
|
665 |
+
gradio_ask,
|
666 |
+
[text_input, chatbot, chat_state, image, img_list, upload_flag, replace_flag],
|
667 |
+
[text_input, chatbot, chat_state, img_list, upload_flag, replace_flag], queue=False
|
668 |
+
).success(
|
669 |
+
gradio_stream_answer,
|
670 |
+
[chatbot, chat_state, img_list, temperature],
|
671 |
+
[chatbot, chat_state]
|
672 |
+
).success(
|
673 |
+
gradio_visualize,
|
674 |
+
[chatbot, image],
|
675 |
+
[chatbot],
|
676 |
+
queue=False,
|
677 |
+
)
|
678 |
+
|
679 |
+
send.click(
|
680 |
+
gradio_ask,
|
681 |
+
[text_input, chatbot, chat_state, image, img_list, upload_flag, replace_flag],
|
682 |
+
[text_input, chatbot, chat_state, img_list, upload_flag, replace_flag], queue=False
|
683 |
+
).success(
|
684 |
+
gradio_stream_answer,
|
685 |
+
[chatbot, chat_state, img_list, temperature],
|
686 |
+
[chatbot, chat_state]
|
687 |
+
).success(
|
688 |
+
gradio_visualize,
|
689 |
+
[chatbot, image],
|
690 |
+
[chatbot],
|
691 |
+
queue=False,
|
692 |
+
)
|
693 |
+
|
694 |
+
clear.click(gradio_reset, [chat_state, img_list], [chatbot, image, text_input, chat_state, img_list], queue=False)
|
695 |
|
696 |
+
demo.launch(share=True, enable_queue=True)
|
|