Spaces:
Paused
Paused
updat
Browse files
app.py
CHANGED
@@ -19,8 +19,8 @@ import torchvision
|
|
19 |
from huggingface_hub import HfApi, login, snapshot_download
|
20 |
from PIL import Image
|
21 |
|
22 |
-
session_token = os.environ.get("SessionToken")
|
23 |
-
login(token=session_token)
|
24 |
|
25 |
csv.field_size_limit(sys.maxsize)
|
26 |
|
@@ -100,22 +100,24 @@ def generate_dataset(username):
|
|
100 |
|
101 |
NUMBER_OF_IMAGES = len(bad_items)
|
102 |
|
|
|
|
|
|
|
103 |
if NUMBER_OF_IMAGES == 0:
|
104 |
return []
|
105 |
|
106 |
-
random_indices = remaining
|
107 |
-
random_images = [imagenet_hard[int(i)]["image"] for i in random_indices]
|
108 |
-
random_gt_ids = [imagenet_hard[int(i)]["label"] for i in random_indices]
|
109 |
-
random_gt_labels = [imagenet_hard[int(x)]["english_label"] for x in random_indices]
|
110 |
|
111 |
data = []
|
112 |
-
for i, image in enumerate(
|
113 |
data.append(
|
114 |
{
|
115 |
-
"id":
|
116 |
-
"
|
117 |
-
"
|
118 |
-
"original_id": int(random_indices[i]),
|
119 |
}
|
120 |
)
|
121 |
return data
|
@@ -153,16 +155,22 @@ def get_training_samples(qid):
|
|
153 |
|
154 |
def load_sample(data, current_index):
|
155 |
image_id = data[current_index]["id"]
|
156 |
-
qimage =
|
|
|
|
|
|
|
|
|
157 |
|
158 |
-
labels = data[current_index]["correct_label"]
|
159 |
return qimage, labels
|
160 |
|
161 |
|
162 |
def preprocessing(data, current_index, history, username):
|
163 |
data = generate_dataset(username)
|
164 |
|
165 |
-
|
|
|
|
|
|
|
166 |
fake_plot = string_to_image("No more images to review")
|
167 |
empty_image = Image.new("RGB", (224, 224))
|
168 |
return (
|
@@ -172,6 +180,7 @@ def preprocessing(data, current_index, history, username):
|
|
172 |
history,
|
173 |
data,
|
174 |
None,
|
|
|
175 |
)
|
176 |
|
177 |
current_index = 0
|
@@ -186,7 +195,15 @@ def preprocessing(data, current_index, history, username):
|
|
186 |
labels = ", ".join(labels)
|
187 |
label_plot = string_to_image(labels)
|
188 |
|
189 |
-
return
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
190 |
|
191 |
|
192 |
def update_app(decision, data, current_index, history, username):
|
@@ -194,7 +211,7 @@ def update_app(decision, data, current_index, history, username):
|
|
194 |
if current_index == -1:
|
195 |
fake_plot = string_to_image("Please Enter your username and load samples")
|
196 |
empty_image = Image.new("RGB", (224, 224))
|
197 |
-
return empty_image, fake_plot, current_index, history, data, None
|
198 |
|
199 |
if current_index == NUMBER_OF_IMAGES - 1:
|
200 |
time_stamp = int(time.time())
|
@@ -226,7 +243,19 @@ def update_app(decision, data, current_index, history, username):
|
|
226 |
|
227 |
fake_plot = string_to_image("Thank you for your time!")
|
228 |
empty_image = Image.new("RGB", (224, 224))
|
229 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
230 |
|
231 |
if current_index >= 0 and current_index < NUMBER_OF_IMAGES - 1:
|
232 |
time_stamp = int(time.time())
|
@@ -270,7 +299,18 @@ def update_app(decision, data, current_index, history, username):
|
|
270 |
labels = ", ".join(labels)
|
271 |
label_plot = string_to_image(labels)
|
272 |
|
273 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
274 |
|
275 |
|
276 |
newcss = """
|
@@ -313,7 +353,11 @@ with gr.Blocks(css=newcss, theme=gr.themes.Soft()) as demo:
|
|
313 |
)
|
314 |
|
315 |
with gr.Column():
|
316 |
-
|
|
|
|
|
|
|
|
|
317 |
prepare_btn = gr.Button(value="Load Samples")
|
318 |
|
319 |
with gr.Column():
|
@@ -341,6 +385,7 @@ with gr.Blocks(css=newcss, theme=gr.themes.Soft()) as demo:
|
|
341 |
history,
|
342 |
data_gr,
|
343 |
training_samples,
|
|
|
344 |
],
|
345 |
)
|
346 |
myabe_btn.click(
|
@@ -353,6 +398,7 @@ with gr.Blocks(css=newcss, theme=gr.themes.Soft()) as demo:
|
|
353 |
history,
|
354 |
data_gr,
|
355 |
training_samples,
|
|
|
356 |
],
|
357 |
)
|
358 |
|
@@ -366,6 +412,7 @@ with gr.Blocks(css=newcss, theme=gr.themes.Soft()) as demo:
|
|
366 |
history,
|
367 |
data_gr,
|
368 |
training_samples,
|
|
|
369 |
],
|
370 |
)
|
371 |
|
@@ -379,7 +426,8 @@ with gr.Blocks(css=newcss, theme=gr.themes.Soft()) as demo:
|
|
379 |
history,
|
380 |
data_gr,
|
381 |
training_samples,
|
|
|
382 |
],
|
383 |
)
|
384 |
|
385 |
-
demo.launch()
|
|
|
19 |
from huggingface_hub import HfApi, login, snapshot_download
|
20 |
from PIL import Image
|
21 |
|
22 |
+
# session_token = os.environ.get("SessionToken")
|
23 |
+
# login(token=session_token)
|
24 |
|
25 |
csv.field_size_limit(sys.maxsize)
|
26 |
|
|
|
100 |
|
101 |
NUMBER_OF_IMAGES = len(bad_items)
|
102 |
|
103 |
+
print(f"NUMBER_OF_IMAGES: {NUMBER_OF_IMAGES}")
|
104 |
+
print(f"Remaining: {len(remaining)}")
|
105 |
+
|
106 |
if NUMBER_OF_IMAGES == 0:
|
107 |
return []
|
108 |
|
109 |
+
# random_indices = remaining
|
110 |
+
# random_images = [imagenet_hard[int(i)]["image"] for i in random_indices]
|
111 |
+
# random_gt_ids = [imagenet_hard[int(i)]["label"] for i in random_indices]
|
112 |
+
# random_gt_labels = [imagenet_hard[int(x)]["english_label"] for x in random_indices]
|
113 |
|
114 |
data = []
|
115 |
+
for i, image in enumerate(remaining):
|
116 |
data.append(
|
117 |
{
|
118 |
+
"id": remaining[i],
|
119 |
+
# "correct_label": random_gt_labels[i],
|
120 |
+
# "original_id": int(random_indices[i]),
|
|
|
121 |
}
|
122 |
)
|
123 |
return data
|
|
|
155 |
|
156 |
def load_sample(data, current_index):
|
157 |
image_id = data[current_index]["id"]
|
158 |
+
qimage = imagenet_hard[int(image_id)]["image"]
|
159 |
+
# labels = data[current_index]["correct_label"]
|
160 |
+
labels = imagenet_hard[int(image_id)]["english_label"]
|
161 |
+
# print(f"Image ID: {image_id}")
|
162 |
+
# print(f"Labels: {labels}")
|
163 |
|
|
|
164 |
return qimage, labels
|
165 |
|
166 |
|
167 |
def preprocessing(data, current_index, history, username):
|
168 |
data = generate_dataset(username)
|
169 |
|
170 |
+
remaining_images = len(data)
|
171 |
+
labeled_images = len(bad_items) - remaining_images
|
172 |
+
|
173 |
+
if remaining_images == 0:
|
174 |
fake_plot = string_to_image("No more images to review")
|
175 |
empty_image = Image.new("RGB", (224, 224))
|
176 |
return (
|
|
|
180 |
history,
|
181 |
data,
|
182 |
None,
|
183 |
+
labeled_images,
|
184 |
)
|
185 |
|
186 |
current_index = 0
|
|
|
195 |
labels = ", ".join(labels)
|
196 |
label_plot = string_to_image(labels)
|
197 |
|
198 |
+
return (
|
199 |
+
qimage,
|
200 |
+
label_plot,
|
201 |
+
current_index,
|
202 |
+
history,
|
203 |
+
data,
|
204 |
+
training_samples_image,
|
205 |
+
labeled_images,
|
206 |
+
)
|
207 |
|
208 |
|
209 |
def update_app(decision, data, current_index, history, username):
|
|
|
211 |
if current_index == -1:
|
212 |
fake_plot = string_to_image("Please Enter your username and load samples")
|
213 |
empty_image = Image.new("RGB", (224, 224))
|
214 |
+
return empty_image, fake_plot, current_index, history, data, None, 0
|
215 |
|
216 |
if current_index == NUMBER_OF_IMAGES - 1:
|
217 |
time_stamp = int(time.time())
|
|
|
243 |
|
244 |
fake_plot = string_to_image("Thank you for your time!")
|
245 |
empty_image = Image.new("RGB", (224, 224))
|
246 |
+
|
247 |
+
remaining_images = len(data)
|
248 |
+
labeled_images = (len(bad_items) - remaining_images) + current_index
|
249 |
+
|
250 |
+
return (
|
251 |
+
empty_image,
|
252 |
+
fake_plot,
|
253 |
+
current_index,
|
254 |
+
history,
|
255 |
+
data,
|
256 |
+
None,
|
257 |
+
labeled_images + 1,
|
258 |
+
)
|
259 |
|
260 |
if current_index >= 0 and current_index < NUMBER_OF_IMAGES - 1:
|
261 |
time_stamp = int(time.time())
|
|
|
299 |
labels = ", ".join(labels)
|
300 |
label_plot = string_to_image(labels)
|
301 |
|
302 |
+
remaining_images = len(data)
|
303 |
+
labeled_images = (len(bad_items) - remaining_images) + current_index
|
304 |
+
|
305 |
+
return (
|
306 |
+
qimage,
|
307 |
+
label_plot,
|
308 |
+
current_index,
|
309 |
+
history,
|
310 |
+
data,
|
311 |
+
training_samples_image,
|
312 |
+
labeled_images,
|
313 |
+
)
|
314 |
|
315 |
|
316 |
newcss = """
|
|
|
353 |
)
|
354 |
|
355 |
with gr.Column():
|
356 |
+
with gr.Row():
|
357 |
+
username = gr.Textbox(label="Username", value=f"user-{random_str}")
|
358 |
+
labeled_images = gr.Textbox(label="Labeled Images", value="0")
|
359 |
+
total_images = gr.Textbox(label="Total Images", value=len(bad_items))
|
360 |
+
|
361 |
prepare_btn = gr.Button(value="Load Samples")
|
362 |
|
363 |
with gr.Column():
|
|
|
385 |
history,
|
386 |
data_gr,
|
387 |
training_samples,
|
388 |
+
labeled_images,
|
389 |
],
|
390 |
)
|
391 |
myabe_btn.click(
|
|
|
398 |
history,
|
399 |
data_gr,
|
400 |
training_samples,
|
401 |
+
labeled_images,
|
402 |
],
|
403 |
)
|
404 |
|
|
|
412 |
history,
|
413 |
data_gr,
|
414 |
training_samples,
|
415 |
+
labeled_images,
|
416 |
],
|
417 |
)
|
418 |
|
|
|
426 |
history,
|
427 |
data_gr,
|
428 |
training_samples,
|
429 |
+
labeled_images,
|
430 |
],
|
431 |
)
|
432 |
|
433 |
+
demo.launch(debug=True)
|