Spaces:
Running
Running
Rishi Desai
commited on
Commit
·
49415e1
1
Parent(s):
5d95766
reworking the demo
Browse files- caption.py +2 -1
- demo.py +440 -238
caption.py
CHANGED
@@ -3,6 +3,7 @@ import io
|
|
3 |
import os
|
4 |
from together import Together
|
5 |
|
|
|
6 |
TRIGGER_WORD = "tr1gger"
|
7 |
|
8 |
def get_system_prompt():
|
@@ -140,7 +141,7 @@ def caption_image_batch(client, image_strings, category):
|
|
140 |
{"role": "user", "content": content}
|
141 |
]
|
142 |
response = client.chat.completions.create(
|
143 |
-
model=
|
144 |
messages=messages
|
145 |
)
|
146 |
return process_batch_response(response, image_strings)
|
|
|
3 |
import os
|
4 |
from together import Together
|
5 |
|
6 |
+
MODEL = "meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8"
|
7 |
TRIGGER_WORD = "tr1gger"
|
8 |
|
9 |
def get_system_prompt():
|
|
|
141 |
{"role": "user", "content": content}
|
142 |
]
|
143 |
response = client.chat.completions.create(
|
144 |
+
model=MODEL,
|
145 |
messages=messages
|
146 |
)
|
147 |
return process_batch_response(response, image_strings)
|
demo.py
CHANGED
@@ -4,12 +4,16 @@ import zipfile
|
|
4 |
from io import BytesIO
|
5 |
import time
|
6 |
import tempfile
|
7 |
-
from main import collect_images_by_category
|
8 |
from pathlib import Path
|
9 |
from caption import caption_images
|
|
|
|
|
10 |
# Maximum number of images
|
11 |
MAX_IMAGES = 30
|
12 |
|
|
|
|
|
13 |
def create_download_file(image_paths, captions):
|
14 |
"""Create a zip file with images and their captions"""
|
15 |
zip_io = BytesIO()
|
@@ -29,129 +33,362 @@ def create_download_file(image_paths, captions):
|
|
29 |
|
30 |
return zip_io.getvalue()
|
31 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
32 |
def process_uploaded_images(image_paths, batch_by_category=False):
|
33 |
"""Process uploaded images using the same code path as CLI"""
|
34 |
try:
|
|
|
|
|
35 |
print(f"Processing {len(image_paths)} images, batch_by_category={batch_by_category}")
|
36 |
-
|
|
|
37 |
with tempfile.TemporaryDirectory() as temp_dir:
|
38 |
-
#
|
39 |
-
temp_image_paths =
|
40 |
-
original_to_temp = {} # Map original paths to temp paths
|
41 |
-
for path in image_paths:
|
42 |
-
filename = os.path.basename(path)
|
43 |
-
temp_path = os.path.join(temp_dir, filename)
|
44 |
-
with open(path, 'rb') as src, open(temp_path, 'wb') as dst:
|
45 |
-
dst.write(src.read())
|
46 |
-
temp_image_paths.append(temp_path)
|
47 |
-
original_to_temp[path] = temp_path
|
48 |
-
|
49 |
-
print(f"Created {len(temp_image_paths)} temporary files")
|
50 |
|
51 |
-
#
|
52 |
temp_dir_path = Path(temp_dir)
|
53 |
|
54 |
-
#
|
|
|
|
|
|
|
|
|
|
|
55 |
images_by_category, image_paths_by_category = collect_images_by_category(temp_dir_path)
|
|
|
|
|
56 |
print(f"Collected images into {len(images_by_category)} categories")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
57 |
|
58 |
-
#
|
59 |
all_images = []
|
60 |
all_image_paths = []
|
61 |
-
|
62 |
-
|
63 |
-
|
64 |
-
|
65 |
-
|
66 |
-
|
67 |
-
|
68 |
-
|
69 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
70 |
break
|
71 |
-
if not found:
|
72 |
-
print(f"Warning: Could not find image {path} in categorized data")
|
73 |
|
74 |
-
print(f"Collected {len(
|
75 |
|
76 |
# Process based on batch setting
|
77 |
-
if batch_by_category:
|
78 |
-
|
79 |
-
captions = [""] * len(image_paths) # Initialize with empty strings
|
80 |
-
for category, images in images_by_category.items():
|
81 |
-
category_paths = image_paths_by_category[category]
|
82 |
-
print(f"Processing category '{category}' with {len(images)} images")
|
83 |
-
# Use the same code path as CLI
|
84 |
-
category_captions = caption_images(images, category=category, batch_mode=True)
|
85 |
-
print(f"Generated {len(category_captions)} captions for category '{category}'")
|
86 |
-
print("Category captions:", category_captions) # Debug print category captions
|
87 |
-
|
88 |
-
# Map captions back to original paths
|
89 |
-
for temp_path, caption in zip(category_paths, category_captions):
|
90 |
-
temp_path_str = str(temp_path)
|
91 |
-
for orig_path, orig_temp in original_to_temp.items():
|
92 |
-
if orig_temp == temp_path_str:
|
93 |
-
idx = image_paths.index(orig_path)
|
94 |
-
captions[idx] = caption
|
95 |
-
break
|
96 |
else:
|
97 |
-
|
98 |
-
|
99 |
-
print(f"Generated {len(all_captions)} captions")
|
100 |
-
print("All captions:", all_captions) # Debug print all captions
|
101 |
-
captions = [""] * len(image_paths)
|
102 |
-
for path, caption in zip(all_image_paths, all_captions):
|
103 |
-
idx = image_paths.index(path)
|
104 |
-
captions[idx] = caption
|
105 |
|
106 |
print(f"Returning {len(captions)} captions")
|
107 |
-
print("Final captions:", captions) # Debug print final captions
|
108 |
return captions
|
109 |
|
110 |
except Exception as e:
|
111 |
print(f"Error in processing: {e}")
|
112 |
raise
|
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 |
<style>
|
156 |
/* Unified styling for the two-column layout */
|
157 |
#left-column, #right-column {
|
@@ -218,10 +455,68 @@ with gr.Blocks() as demo:
|
|
218 |
.download-section {
|
219 |
margin-top: 10px;
|
220 |
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
221 |
</style>
|
222 |
-
"""
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
223 |
|
224 |
-
|
|
|
|
|
|
|
225 |
with gr.Column(visible=False) as captioning_area:
|
226 |
# Replace the single heading with a row containing two headings
|
227 |
with gr.Row():
|
@@ -260,123 +555,15 @@ with gr.Blocks() as demo:
|
|
260 |
)
|
261 |
caption_components.append(caption)
|
262 |
|
263 |
-
|
264 |
-
|
265 |
-
|
266 |
-
|
267 |
-
|
268 |
-
|
269 |
-
|
270 |
-
|
271 |
-
|
272 |
-
gr.Warning(f"Please upload at least one image")
|
273 |
-
return [], gr.update(visible=False), gr.update(interactive=False), gr.update(interactive=False), gr.update(visible=False), gr.update(value="No valid images found"), *[gr.update(visible=False) for _ in range(MAX_IMAGES)]
|
274 |
-
|
275 |
-
if len(image_paths) > MAX_IMAGES:
|
276 |
-
gr.Warning(f"Only the first {MAX_IMAGES} images will be processed")
|
277 |
-
image_paths = image_paths[:MAX_IMAGES]
|
278 |
-
|
279 |
-
# Update row visibility
|
280 |
-
row_updates = []
|
281 |
-
for i in range(MAX_IMAGES):
|
282 |
-
if i < len(image_paths):
|
283 |
-
row_updates.append(gr.update(visible=True))
|
284 |
-
else:
|
285 |
-
row_updates.append(gr.update(visible=False))
|
286 |
-
|
287 |
-
return (
|
288 |
-
image_paths, # stored_image_paths
|
289 |
-
gr.update(visible=True), # captioning_area
|
290 |
-
gr.update(interactive=True), # caption_btn
|
291 |
-
gr.update(interactive=False), # download_btn - initially disabled until captioning is done
|
292 |
-
gr.update(visible=False), # download_output
|
293 |
-
gr.update(value=f"{len(image_paths)} images ready for captioning"), # status_text
|
294 |
-
*row_updates # image_rows
|
295 |
-
)
|
296 |
-
|
297 |
-
def update_images(image_paths):
|
298 |
-
"""Update the image components with the uploaded images"""
|
299 |
-
print(f"Updating images with paths: {image_paths}")
|
300 |
-
updates = []
|
301 |
-
for i in range(MAX_IMAGES):
|
302 |
-
if i < len(image_paths):
|
303 |
-
updates.append(gr.update(value=image_paths[i]))
|
304 |
-
else:
|
305 |
-
updates.append(gr.update(value=None))
|
306 |
-
return updates
|
307 |
-
|
308 |
-
def update_caption_labels(image_paths):
|
309 |
-
"""Update caption labels to include the image filename"""
|
310 |
-
updates = []
|
311 |
-
for i in range(MAX_IMAGES):
|
312 |
-
if i < len(image_paths):
|
313 |
-
filename = os.path.basename(image_paths[i])
|
314 |
-
updates.append(gr.update(label=filename))
|
315 |
-
else:
|
316 |
-
updates.append(gr.update(label=""))
|
317 |
-
return updates
|
318 |
-
|
319 |
-
def run_captioning(image_paths, batch_category):
|
320 |
-
"""Generate captions for the images using the CLI code path"""
|
321 |
-
if not image_paths:
|
322 |
-
return [gr.update(value="") for _ in range(MAX_IMAGES)] + [gr.update(value="No images to process")]
|
323 |
-
|
324 |
-
try:
|
325 |
-
print(f"Starting captioning for {len(image_paths)} images")
|
326 |
-
captions = process_uploaded_images(image_paths, batch_category)
|
327 |
-
print(f"Generated {len(captions)} captions")
|
328 |
-
print("Sample captions:", captions[:2]) # Debug print first two captions
|
329 |
-
|
330 |
-
gr.Info("Captioning complete!")
|
331 |
-
status = gr.update(value="✅ Captioning complete")
|
332 |
-
except Exception as e:
|
333 |
-
print(f"Error in captioning: {str(e)}")
|
334 |
-
gr.Error(f"Captioning failed: {str(e)}")
|
335 |
-
captions = [f"Error: {str(e)}" for _ in image_paths]
|
336 |
-
status = gr.update(value=f"❌ Error: {str(e)}")
|
337 |
-
|
338 |
-
# Update caption textboxes
|
339 |
-
caption_updates = []
|
340 |
-
for i in range(MAX_IMAGES):
|
341 |
-
if i < len(captions):
|
342 |
-
caption_updates.append(gr.update(value=captions[i]))
|
343 |
-
else:
|
344 |
-
caption_updates.append(gr.update(value=""))
|
345 |
-
|
346 |
-
print(f"Returning {len(caption_updates)} caption updates")
|
347 |
-
return caption_updates + [status]
|
348 |
-
|
349 |
-
def update_batch_setting(value):
|
350 |
-
"""Update the batch by category setting"""
|
351 |
-
return value
|
352 |
-
|
353 |
-
def create_zip_from_ui(image_paths, *captions_list):
|
354 |
-
"""Create a zip file from the current images and captions in the UI"""
|
355 |
-
# Filter out empty captions for non-existent images
|
356 |
-
valid_captions = [cap for i, cap in enumerate(captions_list) if i < len(image_paths) and cap]
|
357 |
-
valid_image_paths = image_paths[:len(valid_captions)]
|
358 |
-
|
359 |
-
if not valid_image_paths:
|
360 |
-
gr.Warning("No images to download")
|
361 |
-
return None
|
362 |
-
|
363 |
-
# Create zip file
|
364 |
-
zip_data = create_download_file(valid_image_paths, valid_captions)
|
365 |
-
timestamp = time.strftime("%Y%m%d_%H%M%S")
|
366 |
-
|
367 |
-
# Create a temporary file to store the zip
|
368 |
-
temp_dir = tempfile.gettempdir()
|
369 |
-
zip_filename = f"image_captions_{timestamp}.zip"
|
370 |
-
zip_path = os.path.join(temp_dir, zip_filename)
|
371 |
-
|
372 |
-
# Write the zip data to the temporary file
|
373 |
-
with open(zip_path, "wb") as f:
|
374 |
-
f.write(zip_data)
|
375 |
-
|
376 |
-
# Return the path to the temporary file
|
377 |
-
return zip_path
|
378 |
-
|
379 |
-
# Update the upload_outputs
|
380 |
upload_outputs = [
|
381 |
stored_image_paths,
|
382 |
captioning_area,
|
@@ -387,25 +574,11 @@ with gr.Blocks() as demo:
|
|
387 |
*image_rows
|
388 |
]
|
389 |
|
390 |
-
# Update both paths and images in a single flow
|
391 |
-
def process_upload(files):
|
392 |
-
# First get paths and visibility updates
|
393 |
-
image_paths, captioning_update, caption_btn_update, download_btn_update, download_output_update, status_update, *row_updates = load_captioning(files)
|
394 |
-
|
395 |
-
# Then get image updates
|
396 |
-
image_updates = update_images(image_paths)
|
397 |
-
|
398 |
-
# Update caption labels with filenames
|
399 |
-
caption_label_updates = update_caption_labels(image_paths)
|
400 |
-
|
401 |
-
# Return all updates together
|
402 |
-
return [image_paths, captioning_update, caption_btn_update, download_btn_update, download_output_update, status_update] + row_updates + image_updates + caption_label_updates
|
403 |
-
|
404 |
-
# Combined outputs for both functions
|
405 |
combined_outputs = upload_outputs + image_components + caption_components
|
406 |
|
|
|
407 |
image_upload.change(
|
408 |
-
process_upload,
|
409 |
inputs=[image_upload],
|
410 |
outputs=combined_outputs
|
411 |
)
|
@@ -417,13 +590,6 @@ with gr.Blocks() as demo:
|
|
417 |
outputs=[batch_by_category]
|
418 |
)
|
419 |
|
420 |
-
# Manage the captioning status
|
421 |
-
def on_captioning_start():
|
422 |
-
return gr.update(value="⏳ Processing captions... please wait"), gr.update(interactive=False)
|
423 |
-
|
424 |
-
def on_captioning_complete():
|
425 |
-
return gr.update(value="✅ Captioning complete"), gr.update(interactive=True), gr.update(interactive=True)
|
426 |
-
|
427 |
# Set up captioning button
|
428 |
caption_btn.click(
|
429 |
on_captioning_start,
|
@@ -454,5 +620,41 @@ with gr.Blocks() as demo:
|
|
454 |
outputs=None
|
455 |
)
|
456 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
457 |
if __name__ == "__main__":
|
|
|
458 |
demo.launch(share=True)
|
|
|
4 |
from io import BytesIO
|
5 |
import time
|
6 |
import tempfile
|
7 |
+
from main import collect_images_by_category, is_image_file
|
8 |
from pathlib import Path
|
9 |
from caption import caption_images
|
10 |
+
from PIL import Image
|
11 |
+
|
12 |
# Maximum number of images
|
13 |
MAX_IMAGES = 30
|
14 |
|
15 |
+
# ------- File Operations -------
|
16 |
+
|
17 |
def create_download_file(image_paths, captions):
|
18 |
"""Create a zip file with images and their captions"""
|
19 |
zip_io = BytesIO()
|
|
|
33 |
|
34 |
return zip_io.getvalue()
|
35 |
|
36 |
+
def save_images_to_temp(image_paths, temp_dir):
|
37 |
+
"""Copy images to temporary directory and return mapping"""
|
38 |
+
temp_image_paths = []
|
39 |
+
original_to_temp = {} # Map original paths to temp paths
|
40 |
+
|
41 |
+
for path in image_paths:
|
42 |
+
# Keep original filename to preserve categorization
|
43 |
+
filename = os.path.basename(path)
|
44 |
+
temp_path = os.path.join(temp_dir, filename)
|
45 |
+
|
46 |
+
# Ensure we're using consistent path formats
|
47 |
+
orig_path_str = str(path)
|
48 |
+
temp_path_str = str(temp_path)
|
49 |
+
|
50 |
+
with open(path, 'rb') as src, open(temp_path, 'wb') as dst:
|
51 |
+
dst.write(src.read())
|
52 |
+
|
53 |
+
temp_image_paths.append(temp_path_str)
|
54 |
+
original_to_temp[orig_path_str] = temp_path_str
|
55 |
+
print(f"Copied {orig_path_str} to {temp_path_str}")
|
56 |
+
|
57 |
+
print(f"Created {len(temp_image_paths)} temporary files")
|
58 |
+
return temp_image_paths, original_to_temp
|
59 |
+
|
60 |
+
def process_by_category(images_by_category, image_paths_by_category, image_paths, original_to_temp):
|
61 |
+
"""Process images by category and map captions back to original images"""
|
62 |
+
captions = [""] * len(image_paths) # Initialize with empty strings
|
63 |
+
|
64 |
+
# Create a mapping from temp path to index in the original image_paths
|
65 |
+
temp_to_original_idx = {}
|
66 |
+
for i, orig_path in enumerate(image_paths):
|
67 |
+
if orig_path in original_to_temp:
|
68 |
+
temp_to_original_idx[original_to_temp[orig_path]] = i
|
69 |
+
|
70 |
+
print(f"Created mapping for {len(temp_to_original_idx)} images")
|
71 |
+
|
72 |
+
for category, images in images_by_category.items():
|
73 |
+
category_paths = image_paths_by_category[category]
|
74 |
+
print(f"Processing category '{category}' with {len(images)} images")
|
75 |
+
|
76 |
+
# Create mapping of image to its position in the category
|
77 |
+
category_image_map = {}
|
78 |
+
for i, (img, path) in enumerate(zip(images, category_paths)):
|
79 |
+
category_image_map[str(path)] = i
|
80 |
+
|
81 |
+
try:
|
82 |
+
# Use the same code path as CLI
|
83 |
+
category_captions = caption_images(images, category=category, batch_mode=True)
|
84 |
+
print(f"Generated {len(category_captions)} captions for category '{category}'")
|
85 |
+
|
86 |
+
# Map captions back to original paths using our direct mapping
|
87 |
+
for i, temp_path in enumerate(category_paths):
|
88 |
+
temp_path_str = str(temp_path)
|
89 |
+
if i < len(category_captions) and temp_path_str in temp_to_original_idx:
|
90 |
+
original_idx = temp_to_original_idx[temp_path_str]
|
91 |
+
captions[original_idx] = category_captions[i]
|
92 |
+
except Exception as e:
|
93 |
+
print(f"Error processing category '{category}': {e}")
|
94 |
+
# Fall back to individual processing for this category
|
95 |
+
try:
|
96 |
+
print(f"Falling back to individual processing for category '{category}'")
|
97 |
+
for i, img in enumerate(images):
|
98 |
+
if i >= len(category_paths):
|
99 |
+
continue
|
100 |
+
temp_path = category_paths[i]
|
101 |
+
temp_path_str = str(temp_path)
|
102 |
+
|
103 |
+
try:
|
104 |
+
single_captions = caption_images([img], batch_mode=False)
|
105 |
+
if single_captions and len(single_captions) > 0:
|
106 |
+
if temp_path_str in temp_to_original_idx:
|
107 |
+
original_idx = temp_to_original_idx[temp_path_str]
|
108 |
+
captions[original_idx] = single_captions[0]
|
109 |
+
except Exception as inner_e:
|
110 |
+
print(f"Error processing individual image {i} in '{category}': {inner_e}")
|
111 |
+
except Exception as fallback_e:
|
112 |
+
print(f"Error in fallback processing for '{category}': {fallback_e}")
|
113 |
+
|
114 |
+
return captions
|
115 |
+
|
116 |
+
def process_all_images(all_images, all_image_paths, image_paths):
|
117 |
+
"""Process all images at once without categorization"""
|
118 |
+
print(f"Processing all {len(all_images)} images at once")
|
119 |
+
|
120 |
+
# Initialize empty captions list
|
121 |
+
captions = [""] * len(image_paths) # Initialize with empty strings for all original paths
|
122 |
+
|
123 |
+
# If there are no images, return empty captions
|
124 |
+
if not all_images:
|
125 |
+
print("No images to process, returning empty captions")
|
126 |
+
return captions
|
127 |
+
|
128 |
+
# Create a mapping from temp paths to original indexes for efficient lookup
|
129 |
+
path_to_idx = {str(path): i for i, path in enumerate(image_paths)}
|
130 |
+
|
131 |
+
try:
|
132 |
+
all_captions = caption_images(all_images, batch_mode=False)
|
133 |
+
print(f"Generated {len(all_captions)} captions")
|
134 |
+
|
135 |
+
# Map captions to the right images using the prepared image_paths
|
136 |
+
for i, (path, caption) in enumerate(zip(all_image_paths, all_captions)):
|
137 |
+
if i < len(all_captions) and path in path_to_idx:
|
138 |
+
idx = path_to_idx[path]
|
139 |
+
captions[idx] = caption
|
140 |
+
except Exception as e:
|
141 |
+
print(f"Error generating captions: {e}")
|
142 |
+
|
143 |
+
return captions
|
144 |
+
|
145 |
def process_uploaded_images(image_paths, batch_by_category=False):
|
146 |
"""Process uploaded images using the same code path as CLI"""
|
147 |
try:
|
148 |
+
# Convert all image paths to strings for consistency
|
149 |
+
image_paths = [str(path) for path in image_paths]
|
150 |
print(f"Processing {len(image_paths)} images, batch_by_category={batch_by_category}")
|
151 |
+
|
152 |
+
# Create temporary directory with images
|
153 |
with tempfile.TemporaryDirectory() as temp_dir:
|
154 |
+
# Save images to temp directory
|
155 |
+
temp_image_paths, original_to_temp = save_images_to_temp(image_paths, temp_dir)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
156 |
|
157 |
+
# Use Path object for consistency with main.py
|
158 |
temp_dir_path = Path(temp_dir)
|
159 |
|
160 |
+
# List files in temp directory for debugging
|
161 |
+
print(f"Files in temp directory {temp_dir}:")
|
162 |
+
for f in temp_dir_path.iterdir():
|
163 |
+
print(f" - {f} (is_file: {f.is_file()}, is_image: {is_image_file(f.name)})")
|
164 |
+
|
165 |
+
# Collect images by category using the function from main.py
|
166 |
images_by_category, image_paths_by_category = collect_images_by_category(temp_dir_path)
|
167 |
+
|
168 |
+
# Print categories and counts for debugging
|
169 |
print(f"Collected images into {len(images_by_category)} categories")
|
170 |
+
for category, images in images_by_category.items():
|
171 |
+
print(f" - Category '{category}': {len(images)} images")
|
172 |
+
|
173 |
+
# Check if we actually have images to process
|
174 |
+
total_images = sum(len(images) for images in images_by_category.values())
|
175 |
+
if total_images == 0:
|
176 |
+
print("No images were properly categorized. Adding all images directly.")
|
177 |
+
# Add all images directly without categorization
|
178 |
+
default_category = "default"
|
179 |
+
images_by_category[default_category] = []
|
180 |
+
image_paths_by_category[default_category] = []
|
181 |
+
|
182 |
+
for path in image_paths:
|
183 |
+
path_str = str(path)
|
184 |
+
try:
|
185 |
+
if path_str in original_to_temp:
|
186 |
+
temp_path = original_to_temp[path_str]
|
187 |
+
temp_path_obj = Path(temp_path)
|
188 |
+
img = Image.open(temp_path).convert("RGB")
|
189 |
+
images_by_category[default_category].append(img)
|
190 |
+
image_paths_by_category[default_category].append(temp_path_obj)
|
191 |
+
except Exception as e:
|
192 |
+
print(f"Error loading image {path}: {e}")
|
193 |
|
194 |
+
# Map back to original paths for consistent ordering
|
195 |
all_images = []
|
196 |
all_image_paths = []
|
197 |
+
|
198 |
+
# Create reverse mapping for lookup
|
199 |
+
temp_to_orig = {v: k for k, v in original_to_temp.items()}
|
200 |
+
|
201 |
+
# Go through each category and map back to original
|
202 |
+
for category in images_by_category:
|
203 |
+
for i, temp_path in enumerate(image_paths_by_category[category]):
|
204 |
+
temp_path_str = str(temp_path)
|
205 |
+
if temp_path_str in temp_to_orig:
|
206 |
+
orig_path = temp_to_orig[temp_path_str]
|
207 |
+
if i < len(images_by_category[category]):
|
208 |
+
all_images.append(images_by_category[category][i])
|
209 |
+
all_image_paths.append(orig_path)
|
210 |
+
|
211 |
+
# Ensure we maintain original order
|
212 |
+
ordered_images = []
|
213 |
+
ordered_paths = []
|
214 |
+
|
215 |
+
for orig_path in image_paths:
|
216 |
+
path_str = str(orig_path)
|
217 |
+
for i, path in enumerate(all_image_paths):
|
218 |
+
if path == path_str and i < len(all_images):
|
219 |
+
ordered_images.append(all_images[i])
|
220 |
+
ordered_paths.append(path)
|
221 |
break
|
|
|
|
|
222 |
|
223 |
+
print(f"Collected {len(ordered_images)} images in correct order")
|
224 |
|
225 |
# Process based on batch setting
|
226 |
+
if batch_by_category and len(images_by_category) > 0:
|
227 |
+
captions = process_by_category(images_by_category, image_paths_by_category, image_paths, original_to_temp)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
228 |
else:
|
229 |
+
# Use our own function for non-batch mode since it needs to map back to UI
|
230 |
+
captions = process_all_images(ordered_images, ordered_paths, image_paths)
|
|
|
|
|
|
|
|
|
|
|
|
|
231 |
|
232 |
print(f"Returning {len(captions)} captions")
|
|
|
233 |
return captions
|
234 |
|
235 |
except Exception as e:
|
236 |
print(f"Error in processing: {e}")
|
237 |
raise
|
238 |
|
239 |
+
# ------- UI Interaction Functions -------
|
240 |
+
|
241 |
+
def load_captioning(files):
|
242 |
+
"""Process uploaded images and show them in the UI"""
|
243 |
+
if not files:
|
244 |
+
return [], gr.update(visible=False), gr.update(interactive=False), gr.update(interactive=False), gr.update(visible=False), gr.update(value="Upload images to begin"), *[gr.update(visible=False) for _ in range(MAX_IMAGES)]
|
245 |
+
|
246 |
+
# Filter to only keep image files
|
247 |
+
image_paths = [f for f in files if f.lower().endswith(('.png', '.jpg', '.jpeg', '.gif', '.bmp', '.webp'))]
|
248 |
+
|
249 |
+
if not image_paths or len(image_paths) < 1:
|
250 |
+
gr.Warning(f"Please upload at least one image")
|
251 |
+
return [], gr.update(visible=False), gr.update(interactive=False), gr.update(interactive=False), gr.update(visible=False), gr.update(value="No valid images found"), *[gr.update(visible=False) for _ in range(MAX_IMAGES)]
|
252 |
+
|
253 |
+
if len(image_paths) > MAX_IMAGES:
|
254 |
+
gr.Warning(f"Only the first {MAX_IMAGES} images will be processed")
|
255 |
+
image_paths = image_paths[:MAX_IMAGES]
|
256 |
+
|
257 |
+
# Update row visibility
|
258 |
+
row_updates = []
|
259 |
+
for i in range(MAX_IMAGES):
|
260 |
+
if i < len(image_paths):
|
261 |
+
row_updates.append(gr.update(visible=True))
|
262 |
+
else:
|
263 |
+
row_updates.append(gr.update(visible=False))
|
264 |
+
|
265 |
+
return (
|
266 |
+
image_paths, # stored_image_paths
|
267 |
+
gr.update(visible=True), # captioning_area
|
268 |
+
gr.update(interactive=True), # caption_btn
|
269 |
+
gr.update(interactive=False), # download_btn - initially disabled until captioning is done
|
270 |
+
gr.update(visible=False), # download_output
|
271 |
+
gr.update(value=f"{len(image_paths)} images ready for captioning"), # status_text
|
272 |
+
*row_updates # image_rows
|
273 |
+
)
|
274 |
+
|
275 |
+
def update_images(image_paths):
|
276 |
+
"""Update the image components with the uploaded images"""
|
277 |
+
print(f"Updating images with paths: {image_paths}")
|
278 |
+
updates = []
|
279 |
+
for i in range(MAX_IMAGES):
|
280 |
+
if i < len(image_paths):
|
281 |
+
updates.append(gr.update(value=image_paths[i]))
|
282 |
+
else:
|
283 |
+
updates.append(gr.update(value=None))
|
284 |
+
return updates
|
285 |
+
|
286 |
+
def update_caption_labels(image_paths):
|
287 |
+
"""Update caption labels to include the image filename"""
|
288 |
+
updates = []
|
289 |
+
for i in range(MAX_IMAGES):
|
290 |
+
if i < len(image_paths):
|
291 |
+
filename = os.path.basename(image_paths[i])
|
292 |
+
updates.append(gr.update(label=filename))
|
293 |
+
else:
|
294 |
+
updates.append(gr.update(label=""))
|
295 |
+
return updates
|
296 |
+
|
297 |
+
def run_captioning(image_paths, batch_category):
|
298 |
+
"""Generate captions for the images using the CLI code path"""
|
299 |
+
if not image_paths:
|
300 |
+
return [gr.update(value="") for _ in range(MAX_IMAGES)] + [gr.update(value="No images to process")]
|
301 |
|
302 |
+
try:
|
303 |
+
print(f"Starting captioning for {len(image_paths)} images, batch_by_category={batch_category}")
|
304 |
+
captions = process_uploaded_images(image_paths, batch_category)
|
305 |
+
|
306 |
+
# Count valid captions
|
307 |
+
valid_captions = sum(1 for c in captions if c and c.strip())
|
308 |
+
print(f"Generated {valid_captions} valid captions out of {len(captions)} images")
|
309 |
+
|
310 |
+
if valid_captions < len(captions):
|
311 |
+
gr.Warning(f"{len(captions) - valid_captions} images could not be captioned properly")
|
312 |
+
status = gr.update(value=f"✅ Captioning complete - {valid_captions}/{len(captions)} successful")
|
313 |
+
else:
|
314 |
+
gr.Info("Captioning complete!")
|
315 |
+
status = gr.update(value="✅ Captioning complete")
|
316 |
+
|
317 |
+
print("Sample captions:", captions[:2] if len(captions) >= 2 else captions)
|
318 |
+
except Exception as e:
|
319 |
+
print(f"Error in captioning: {str(e)}")
|
320 |
+
gr.Error(f"Captioning failed: {str(e)}")
|
321 |
+
captions = [""] * len(image_paths) # Use empty strings
|
322 |
+
status = gr.update(value=f"❌ Error: {str(e)}")
|
323 |
+
|
324 |
+
# Update caption textboxes
|
325 |
+
caption_updates = []
|
326 |
+
for i in range(MAX_IMAGES):
|
327 |
+
if i < len(captions) and captions[i]: # Only set value if we have a valid caption
|
328 |
+
caption_updates.append(gr.update(value=captions[i]))
|
329 |
+
else:
|
330 |
+
caption_updates.append(gr.update(value=""))
|
331 |
+
|
332 |
+
print(f"Returning {len(caption_updates)} caption updates")
|
333 |
+
return caption_updates + [status]
|
334 |
+
|
335 |
+
def update_batch_setting(value):
|
336 |
+
"""Update the batch by category setting"""
|
337 |
+
return value
|
338 |
+
|
339 |
+
def create_zip_from_ui(image_paths, *captions_list):
|
340 |
+
"""Create a zip file from the current images and captions in the UI"""
|
341 |
+
# Filter out empty captions for non-existent images
|
342 |
+
valid_captions = [cap for i, cap in enumerate(captions_list) if i < len(image_paths) and cap]
|
343 |
+
valid_image_paths = image_paths[:len(valid_captions)]
|
344 |
+
|
345 |
+
if not valid_image_paths:
|
346 |
+
gr.Warning("No images to download")
|
347 |
+
return None
|
348 |
+
|
349 |
+
# Create zip file
|
350 |
+
zip_data = create_download_file(valid_image_paths, valid_captions)
|
351 |
+
timestamp = time.strftime("%Y%m%d_%H%M%S")
|
352 |
+
|
353 |
+
# Create a temporary file to store the zip
|
354 |
+
temp_dir = tempfile.gettempdir()
|
355 |
+
zip_filename = f"image_captions_{timestamp}.zip"
|
356 |
+
zip_path = os.path.join(temp_dir, zip_filename)
|
357 |
+
|
358 |
+
# Write the zip data to the temporary file
|
359 |
+
with open(zip_path, "wb") as f:
|
360 |
+
f.write(zip_data)
|
361 |
+
|
362 |
+
# Return the path to the temporary file
|
363 |
+
return zip_path
|
364 |
+
|
365 |
+
def process_upload(files, image_rows, image_components, caption_components):
|
366 |
+
"""Process uploaded files and update UI components"""
|
367 |
+
# First get paths and visibility updates
|
368 |
+
image_paths, captioning_update, caption_btn_update, download_btn_update, download_output_update, status_update, *row_updates = load_captioning(files)
|
369 |
+
|
370 |
+
# Then get image updates
|
371 |
+
image_updates = update_images(image_paths)
|
372 |
|
373 |
+
# Update caption labels with filenames
|
374 |
+
caption_label_updates = update_caption_labels(image_paths)
|
375 |
+
|
376 |
+
# Return all updates together
|
377 |
+
return [image_paths, captioning_update, caption_btn_update, download_btn_update, download_output_update, status_update] + row_updates + image_updates + caption_label_updates
|
378 |
+
|
379 |
+
def on_captioning_start():
|
380 |
+
"""Update UI when captioning starts"""
|
381 |
+
return gr.update(value="⏳ Processing captions... please wait"), gr.update(interactive=False)
|
382 |
+
|
383 |
+
def on_captioning_complete():
|
384 |
+
"""Update UI when captioning completes"""
|
385 |
+
return gr.update(value="✅ Captioning complete"), gr.update(interactive=True), gr.update(interactive=True)
|
386 |
+
|
387 |
+
# ------- UI Style Definitions -------
|
388 |
+
|
389 |
+
def get_css_styles():
|
390 |
+
"""Return CSS styles for the UI"""
|
391 |
+
return """
|
392 |
<style>
|
393 |
/* Unified styling for the two-column layout */
|
394 |
#left-column, #right-column {
|
|
|
455 |
.download-section {
|
456 |
margin-top: 10px;
|
457 |
}
|
458 |
+
|
459 |
+
/* Category info */
|
460 |
+
.category-info {
|
461 |
+
font-size: 0.9em;
|
462 |
+
color: #555;
|
463 |
+
background-color: #f8f9fa;
|
464 |
+
padding: 8px;
|
465 |
+
border-radius: 4px;
|
466 |
+
margin-bottom: 10px;
|
467 |
+
border-left: 3px solid #4CAF50;
|
468 |
+
}
|
469 |
</style>
|
470 |
+
"""
|
471 |
+
|
472 |
+
# ------- UI Component Creation -------
|
473 |
+
|
474 |
+
def create_upload_area():
|
475 |
+
"""Create the upload area components"""
|
476 |
+
# Left column for images/upload
|
477 |
+
with gr.Column(scale=1, elem_id="left-column") as upload_column:
|
478 |
+
# Upload area
|
479 |
+
gr.Markdown("### Upload your images", elem_id="upload-heading")
|
480 |
+
gr.Markdown("Only .png, .jpg, .jpeg, and .webp files are supported", elem_id="file-types-info", elem_classes="file-types-info")
|
481 |
+
image_upload = gr.File(
|
482 |
+
file_count="multiple",
|
483 |
+
label="Drop your files here",
|
484 |
+
file_types=["image"],
|
485 |
+
type="filepath",
|
486 |
+
height=220,
|
487 |
+
elem_classes="file-upload-container",
|
488 |
+
)
|
489 |
+
|
490 |
+
return upload_column, image_upload
|
491 |
+
|
492 |
+
def create_config_area():
|
493 |
+
"""Create the configuration area components"""
|
494 |
+
# Right column for configuration and captions
|
495 |
+
with gr.Column(scale=1.5, elem_id="right-column") as config_column:
|
496 |
+
# Configuration area
|
497 |
+
gr.Markdown("### Configuration")
|
498 |
+
batch_category_checkbox = gr.Checkbox(
|
499 |
+
label="Batch process by category",
|
500 |
+
value=False,
|
501 |
+
info="Caption similar images together"
|
502 |
+
)
|
503 |
+
|
504 |
+
gr.Markdown("""
|
505 |
+
**Note about categorization:**
|
506 |
+
- Images are grouped by the part of the filename before the last underscore
|
507 |
+
- For example: "character_pose_01.png" and "character_pose_02.png" share the category "character_pose"
|
508 |
+
- When using "Batch by category", similar images are captioned together for more consistent results
|
509 |
+
""", elem_classes=["category-info"])
|
510 |
+
|
511 |
+
caption_btn = gr.Button("Caption Images", variant="primary", interactive=False)
|
512 |
+
download_btn = gr.Button("Download Images + Captions", variant="secondary", interactive=False)
|
513 |
+
download_output = gr.File(label="Download Zip", visible=False)
|
514 |
+
status_text = gr.Markdown("Upload images to begin", visible=True)
|
515 |
|
516 |
+
return config_column, batch_category_checkbox, caption_btn, download_btn, download_output, status_text
|
517 |
+
|
518 |
+
def create_captioning_area():
|
519 |
+
"""Create the captioning area components"""
|
520 |
with gr.Column(visible=False) as captioning_area:
|
521 |
# Replace the single heading with a row containing two headings
|
522 |
with gr.Row():
|
|
|
555 |
)
|
556 |
caption_components.append(caption)
|
557 |
|
558 |
+
return captioning_area, image_rows, image_components, caption_components
|
559 |
+
|
560 |
+
def setup_event_handlers(
|
561 |
+
image_upload, stored_image_paths, captioning_area, caption_btn, download_btn,
|
562 |
+
download_output, status_text, image_rows, image_components, caption_components,
|
563 |
+
batch_category_checkbox, batch_by_category
|
564 |
+
):
|
565 |
+
"""Set up all event handlers for the UI"""
|
566 |
+
# Combined outputs for the upload function
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
567 |
upload_outputs = [
|
568 |
stored_image_paths,
|
569 |
captioning_area,
|
|
|
574 |
*image_rows
|
575 |
]
|
576 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
577 |
combined_outputs = upload_outputs + image_components + caption_components
|
578 |
|
579 |
+
# Set up upload handler
|
580 |
image_upload.change(
|
581 |
+
lambda files: process_upload(files, image_rows, image_components, caption_components),
|
582 |
inputs=[image_upload],
|
583 |
outputs=combined_outputs
|
584 |
)
|
|
|
590 |
outputs=[batch_by_category]
|
591 |
)
|
592 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
593 |
# Set up captioning button
|
594 |
caption_btn.click(
|
595 |
on_captioning_start,
|
|
|
620 |
outputs=None
|
621 |
)
|
622 |
|
623 |
+
# ------- Main Application -------
|
624 |
+
|
625 |
+
def build_ui():
|
626 |
+
"""Build and return the Gradio interface"""
|
627 |
+
with gr.Blocks() as demo:
|
628 |
+
gr.Markdown("# Image Auto-captioner for LoRA Training")
|
629 |
+
|
630 |
+
# Store uploaded images
|
631 |
+
stored_image_paths = gr.State([])
|
632 |
+
batch_by_category = gr.State(False) # State to track if batch by category is enabled
|
633 |
+
|
634 |
+
# Create a two-column layout for the entire interface
|
635 |
+
with gr.Row():
|
636 |
+
# Create upload area in left column
|
637 |
+
_, image_upload = create_upload_area()
|
638 |
+
|
639 |
+
# Create config area in right column
|
640 |
+
_, batch_category_checkbox, caption_btn, download_btn, download_output, status_text = create_config_area()
|
641 |
+
|
642 |
+
# Add CSS styling
|
643 |
+
gr.HTML(get_css_styles())
|
644 |
+
|
645 |
+
# Create captioning area (initially hidden)
|
646 |
+
captioning_area, image_rows, image_components, caption_components = create_captioning_area()
|
647 |
+
|
648 |
+
# Set up event handlers
|
649 |
+
setup_event_handlers(
|
650 |
+
image_upload, stored_image_paths, captioning_area, caption_btn, download_btn,
|
651 |
+
download_output, status_text, image_rows, image_components, caption_components,
|
652 |
+
batch_category_checkbox, batch_by_category
|
653 |
+
)
|
654 |
+
|
655 |
+
return demo
|
656 |
+
|
657 |
+
# Launch the app when run directly
|
658 |
if __name__ == "__main__":
|
659 |
+
demo = build_ui()
|
660 |
demo.launch(share=True)
|