Upload 2 files
Browse files- app.py +312 -240
- requirements.txt +1 -1
app.py
CHANGED
@@ -27,7 +27,7 @@ from CLIP_Explainability.vit_cam import (
|
|
27 |
|
28 |
from pytorch_grad_cam.grad_cam import GradCAM
|
29 |
|
30 |
-
RUN_LITE =
|
31 |
|
32 |
MAX_IMG_WIDTH = 500
|
33 |
MAX_IMG_HEIGHT = 800
|
@@ -58,7 +58,10 @@ def encode_search_query(search_query, model_type):
|
|
58 |
text_encoded /= text_encoded.norm(dim=-1, keepdim=True)
|
59 |
elif model_type == "J-CLIP (日本語 ViT)":
|
60 |
t_text = st.session_state.ja_tokenizer(
|
61 |
-
search_query,
|
|
|
|
|
|
|
62 |
)
|
63 |
text_encoded = st.session_state.ja_model.get_text_features(**t_text)
|
64 |
text_encoded /= text_encoded.norm(dim=-1, keepdim=True)
|
@@ -67,7 +70,7 @@ def encode_search_query(search_query, model_type):
|
|
67 |
text_encoded /= text_encoded.norm(dim=-1, keepdim=True)
|
68 |
|
69 |
# Retrieve the feature vector
|
70 |
-
return text_encoded
|
71 |
|
72 |
|
73 |
def clip_search(search_query):
|
@@ -153,7 +156,9 @@ def load_image_features():
|
|
153 |
def init():
|
154 |
st.session_state.current_page = 1
|
155 |
|
156 |
-
device = "cuda" if torch.cuda.is_available() else "cpu"
|
|
|
|
|
157 |
st.session_state.device = device
|
158 |
|
159 |
# Load the open CLIP models
|
@@ -168,7 +173,7 @@ def init():
|
|
168 |
|
169 |
st.session_state.ml_model = (
|
170 |
pt_multilingual_clip.MultilingualCLIP.from_pretrained(ml_model_name)
|
171 |
-
)
|
172 |
st.session_state.ml_tokenizer = AutoTokenizer.from_pretrained(ml_model_name)
|
173 |
|
174 |
ja_model_name = "hakuhodo-tech/japanese-clip-vit-h-14-bert-wider"
|
@@ -193,7 +198,7 @@ def init():
|
|
193 |
|
194 |
st.session_state.rn_model = legacy_multilingual_clip.load_model(
|
195 |
"M-BERT-Base-69"
|
196 |
-
)
|
197 |
st.session_state.rn_tokenizer = BertTokenizer.from_pretrained(
|
198 |
"bert-base-multilingual-cased"
|
199 |
)
|
@@ -210,7 +215,6 @@ def init():
|
|
210 |
st.session_state.vision_mode = "tiled"
|
211 |
st.session_state.search_image_ids = []
|
212 |
st.session_state.search_image_scores = {}
|
213 |
-
st.session_state.activations_image = None
|
214 |
st.session_state.text_table_df = None
|
215 |
|
216 |
with st.spinner("Loading models and data, please wait..."):
|
@@ -221,233 +225,271 @@ if "images_info" not in st.session_state:
|
|
221 |
init()
|
222 |
|
223 |
|
224 |
-
def
|
225 |
-
|
226 |
-
return
|
227 |
|
228 |
-
|
229 |
-
|
230 |
-
st.title("Image + query details")
|
231 |
-
with header_cols[1]:
|
232 |
-
if st.button("Close"):
|
233 |
-
st.rerun()
|
234 |
|
235 |
-
st.
|
236 |
-
|
237 |
-
)
|
238 |
|
239 |
-
|
240 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
241 |
|
242 |
-
|
243 |
-
|
244 |
-
|
245 |
-
|
|
|
246 |
|
247 |
-
|
248 |
-
|
249 |
-
|
250 |
-
|
251 |
-
|
252 |
-
|
253 |
-
|
254 |
-
|
255 |
-
|
256 |
-
|
257 |
-
|
258 |
-
if st.session_state.vision_mode == "tiled":
|
259 |
-
scaled_dims = [img_dim, img_dim]
|
260 |
-
|
261 |
-
if orig_img_dims[0] > orig_img_dims[1]:
|
262 |
-
scale_ratio = round(orig_img_dims[0] / orig_img_dims[1])
|
263 |
-
if scale_ratio > 1:
|
264 |
-
scaled_dims = [scale_ratio * img_dim, img_dim]
|
265 |
-
tile_behavior = "width"
|
266 |
-
elif orig_img_dims[0] < orig_img_dims[1]:
|
267 |
-
scale_ratio = round(orig_img_dims[1] / orig_img_dims[0])
|
268 |
-
if scale_ratio > 1:
|
269 |
-
scaled_dims = [img_dim, scale_ratio * img_dim]
|
270 |
-
tile_behavior = "height"
|
271 |
-
|
272 |
-
resized_image = image.resize(scaled_dims, Image.LANCZOS)
|
273 |
-
|
274 |
-
if tile_behavior == "width":
|
275 |
-
image_tiles = []
|
276 |
-
for x in range(0, scale_ratio):
|
277 |
-
box = (x * img_dim, 0, (x + 1) * img_dim, img_dim)
|
278 |
-
image_tiles.append(resized_image.crop(box))
|
279 |
-
|
280 |
-
elif tile_behavior == "height":
|
281 |
-
image_tiles = []
|
282 |
-
for y in range(0, scale_ratio):
|
283 |
-
box = (0, y * img_dim, img_dim, (y + 1) * img_dim)
|
284 |
-
image_tiles.append(resized_image.crop(box))
|
285 |
-
|
286 |
-
else:
|
287 |
-
image_tiles = [resized_image]
|
288 |
-
|
289 |
-
elif st.session_state.vision_mode == "stretched":
|
290 |
-
image_tiles = [image.resize((img_dim, img_dim), Image.LANCZOS)]
|
291 |
-
|
292 |
-
else: # vision_mode == "cropped"
|
293 |
-
if orig_img_dims[0] > orig_img_dims[1]:
|
294 |
-
scale_factor = orig_img_dims[0] / orig_img_dims[1]
|
295 |
-
resized_img_dims = (round(scale_factor * img_dim), img_dim)
|
296 |
-
resized_img = image.resize(resized_img_dims)
|
297 |
-
elif orig_img_dims[0] < orig_img_dims[1]:
|
298 |
-
scale_factor = orig_img_dims[1] / orig_img_dims[0]
|
299 |
-
resized_img_dims = (img_dim, round(scale_factor * img_dim))
|
300 |
-
else:
|
301 |
-
resized_img_dims = (img_dim, img_dim)
|
302 |
|
|
|
|
|
|
|
|
|
303 |
resized_img = image.resize(resized_img_dims)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
304 |
|
305 |
-
|
306 |
-
|
307 |
-
x_right = round(resized_img_dims[0] - img_dim) - left
|
308 |
-
x_bottom = round(resized_img_dims[1] - img_dim) - top
|
309 |
-
right = resized_img_dims[0] - x_right
|
310 |
-
bottom = resized_img_dims[1] - x_bottom
|
311 |
|
312 |
-
|
313 |
-
|
|
|
314 |
|
315 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
316 |
|
317 |
-
|
318 |
-
|
319 |
-
|
320 |
-
|
|
|
321 |
)
|
322 |
|
323 |
-
|
324 |
-
|
|
|
|
|
|
|
|
|
325 |
)
|
326 |
|
327 |
-
|
328 |
-
|
329 |
-
|
330 |
-
|
331 |
-
|
332 |
-
|
333 |
-
|
334 |
-
|
335 |
-
|
336 |
-
|
337 |
-
|
338 |
-
|
339 |
-
|
340 |
-
|
341 |
-
|
342 |
-
|
343 |
-
|
344 |
-
|
345 |
-
|
346 |
-
|
347 |
-
|
348 |
-
|
349 |
-
st.session_state.
|
350 |
)
|
351 |
|
352 |
-
|
353 |
-
|
|
|
|
|
|
|
|
|
354 |
)
|
355 |
-
|
356 |
-
|
357 |
-
|
358 |
-
|
359 |
-
|
360 |
-
|
361 |
-
|
362 |
-
|
363 |
-
|
364 |
-
|
365 |
-
|
366 |
-
|
367 |
-
|
368 |
-
|
369 |
-
|
370 |
-
|
371 |
-
|
372 |
-
)
|
373 |
-
|
374 |
-
image_visualizations.append(vis_t)
|
375 |
-
|
376 |
-
else: # st.session_state.active_model == Legacy
|
377 |
-
# Sometimes used for token importance viz
|
378 |
-
tokenized_text = st.session_state.rn_tokenizer.tokenize(
|
379 |
-
st.session_state.search_field_value
|
380 |
)
|
381 |
|
382 |
-
|
383 |
-
|
|
|
|
|
|
|
|
|
|
|
384 |
)
|
385 |
|
386 |
-
|
|
|
387 |
|
388 |
-
|
389 |
-
|
390 |
-
|
391 |
-
.unsqueeze(0)
|
392 |
-
.to(st.session_state.device)
|
393 |
-
)
|
394 |
|
395 |
-
|
396 |
-
p_image.type(st.session_state.rn_image_model.dtype),
|
397 |
-
text_features,
|
398 |
-
image_model.visual,
|
399 |
-
GradCAM,
|
400 |
-
st.session_state.device,
|
401 |
-
img_dim=img_dim,
|
402 |
-
)
|
403 |
|
404 |
-
|
|
|
|
|
405 |
|
406 |
-
|
407 |
|
408 |
-
|
409 |
|
410 |
-
|
411 |
-
|
412 |
-
|
413 |
|
414 |
-
|
415 |
-
|
416 |
-
|
417 |
-
|
418 |
-
|
419 |
-
|
420 |
|
421 |
-
|
422 |
-
|
423 |
-
|
424 |
-
|
425 |
-
|
426 |
|
427 |
-
|
428 |
-
|
429 |
-
|
430 |
-
|
431 |
-
|
432 |
|
433 |
-
|
434 |
-
|
435 |
|
436 |
-
|
437 |
-
|
438 |
-
|
439 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
440 |
)
|
441 |
|
442 |
-
|
443 |
-
|
444 |
-
|
445 |
-
|
|
|
446 |
)
|
447 |
|
448 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
449 |
tokenized_text = [
|
450 |
-
tok
|
|
|
|
|
|
|
451 |
]
|
452 |
|
453 |
if (
|
@@ -457,8 +499,7 @@ def visualize_gradcam(viz_image_id):
|
|
457 |
"Calculate text importance (may take some time)",
|
458 |
)
|
459 |
):
|
460 |
-
|
461 |
-
token_scores = []
|
462 |
|
463 |
progress_text = f"Processing {len(tokenized_text)} text tokens"
|
464 |
progress_bar = st.progress(0.0, text=progress_text)
|
@@ -466,34 +507,37 @@ def visualize_gradcam(viz_image_id):
|
|
466 |
for t, tok in enumerate(tokenized_text):
|
467 |
token = tok
|
468 |
|
469 |
-
|
470 |
-
|
471 |
-
|
472 |
-
|
473 |
-
|
474 |
-
|
475 |
-
|
476 |
-
|
477 |
-
|
478 |
-
|
479 |
-
|
480 |
-
|
481 |
-
|
482 |
-
|
483 |
-
|
484 |
-
|
485 |
-
|
486 |
-
|
487 |
-
|
488 |
-
|
489 |
-
|
490 |
-
|
491 |
-
|
492 |
-
|
493 |
-
|
494 |
-
|
495 |
-
|
496 |
-
|
|
|
|
|
|
|
497 |
|
498 |
progress_bar.progress(
|
499 |
(t + 1) / len(tokenized_text),
|
@@ -501,24 +545,48 @@ def visualize_gradcam(viz_image_id):
|
|
501 |
)
|
502 |
progress_bar.empty()
|
503 |
|
504 |
-
|
|
|
|
|
|
|
|
|
505 |
|
506 |
token_scores = [f"{round(score.item() * 100, 3)}%" for score in normed_scores]
|
507 |
st.session_state.text_table_df = pd.DataFrame(
|
508 |
-
{"token":
|
509 |
)
|
510 |
|
511 |
st.markdown("**Importance of each text token to relevance score**")
|
512 |
st.table(st.session_state.text_table_df)
|
513 |
|
514 |
|
515 |
-
|
516 |
-
|
|
|
517 |
|
518 |
|
519 |
-
|
520 |
-
|
521 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
522 |
|
523 |
|
524 |
st.title("Explore Japanese visual aesthetics with CLIP models")
|
@@ -637,7 +705,7 @@ else:
|
|
637 |
use_container_width=True,
|
638 |
)
|
639 |
|
640 |
-
controls = st.columns([
|
641 |
with controls[0]:
|
642 |
im_per_pg = st.columns([30, 70], vertical_alignment="center")
|
643 |
with im_per_pg[0]:
|
@@ -647,8 +715,6 @@ with controls[0]:
|
|
647 |
"Images/page:", range(10, 50, 10), label_visibility="collapsed"
|
648 |
)
|
649 |
with controls[1]:
|
650 |
-
st.empty()
|
651 |
-
with controls[2]:
|
652 |
im_per_row = st.columns([30, 70], vertical_alignment="center")
|
653 |
with im_per_row[0]:
|
654 |
st.markdown("**Images/row:**")
|
@@ -657,9 +723,7 @@ with controls[2]:
|
|
657 |
"Images/row:", range(1, 6), value=5, label_visibility="collapsed"
|
658 |
)
|
659 |
num_batches = ceil(len(st.session_state.image_ids) / batch_size)
|
660 |
-
with controls[
|
661 |
-
st.empty()
|
662 |
-
with controls[4]:
|
663 |
pager = st.columns([40, 60], vertical_alignment="center")
|
664 |
with pager[0]:
|
665 |
st.markdown(f"Page **{st.session_state.current_page}** of **{num_batches}** ")
|
@@ -672,6 +736,14 @@ with controls[4]:
|
|
672 |
label_visibility="collapsed",
|
673 |
key="current_page",
|
674 |
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
675 |
|
676 |
|
677 |
if len(st.session_state.search_image_ids) == 0:
|
@@ -708,7 +780,7 @@ for image_id in batch:
|
|
708 |
if not RUN_LITE or st.session_state.active_model == "M-CLIP (multilingual ViT)":
|
709 |
st.button(
|
710 |
"Explain this",
|
711 |
-
on_click=
|
712 |
args=[image_id],
|
713 |
use_container_width=True,
|
714 |
key=image_id,
|
|
|
27 |
|
28 |
from pytorch_grad_cam.grad_cam import GradCAM
|
29 |
|
30 |
+
RUN_LITE = True # Load vision model for CAM viz explainability for M-CLIP only
|
31 |
|
32 |
MAX_IMG_WIDTH = 500
|
33 |
MAX_IMG_HEIGHT = 800
|
|
|
58 |
text_encoded /= text_encoded.norm(dim=-1, keepdim=True)
|
59 |
elif model_type == "J-CLIP (日本語 ViT)":
|
60 |
t_text = st.session_state.ja_tokenizer(
|
61 |
+
search_query,
|
62 |
+
padding=True,
|
63 |
+
return_tensors="pt",
|
64 |
+
device=st.session_state.device,
|
65 |
)
|
66 |
text_encoded = st.session_state.ja_model.get_text_features(**t_text)
|
67 |
text_encoded /= text_encoded.norm(dim=-1, keepdim=True)
|
|
|
70 |
text_encoded /= text_encoded.norm(dim=-1, keepdim=True)
|
71 |
|
72 |
# Retrieve the feature vector
|
73 |
+
return text_encoded.to(st.session_state.device)
|
74 |
|
75 |
|
76 |
def clip_search(search_query):
|
|
|
156 |
def init():
|
157 |
st.session_state.current_page = 1
|
158 |
|
159 |
+
# device = "cuda" if torch.cuda.is_available() else "cpu"
|
160 |
+
device = "cpu"
|
161 |
+
|
162 |
st.session_state.device = device
|
163 |
|
164 |
# Load the open CLIP models
|
|
|
173 |
|
174 |
st.session_state.ml_model = (
|
175 |
pt_multilingual_clip.MultilingualCLIP.from_pretrained(ml_model_name)
|
176 |
+
).to(device)
|
177 |
st.session_state.ml_tokenizer = AutoTokenizer.from_pretrained(ml_model_name)
|
178 |
|
179 |
ja_model_name = "hakuhodo-tech/japanese-clip-vit-h-14-bert-wider"
|
|
|
198 |
|
199 |
st.session_state.rn_model = legacy_multilingual_clip.load_model(
|
200 |
"M-BERT-Base-69"
|
201 |
+
).to(device)
|
202 |
st.session_state.rn_tokenizer = BertTokenizer.from_pretrained(
|
203 |
"bert-base-multilingual-cased"
|
204 |
)
|
|
|
215 |
st.session_state.vision_mode = "tiled"
|
216 |
st.session_state.search_image_ids = []
|
217 |
st.session_state.search_image_scores = {}
|
|
|
218 |
st.session_state.text_table_df = None
|
219 |
|
220 |
with st.spinner("Loading models and data, please wait..."):
|
|
|
225 |
init()
|
226 |
|
227 |
|
228 |
+
def get_overlay_vis(image, img_dim, image_model):
|
229 |
+
orig_img_dims = image.size
|
|
|
230 |
|
231 |
+
##### If the features are based on tiled image slices
|
232 |
+
tile_behavior = None
|
|
|
|
|
|
|
|
|
233 |
|
234 |
+
if st.session_state.vision_mode == "tiled":
|
235 |
+
scaled_dims = [img_dim, img_dim]
|
|
|
236 |
|
237 |
+
if orig_img_dims[0] > orig_img_dims[1]:
|
238 |
+
scale_ratio = round(orig_img_dims[0] / orig_img_dims[1])
|
239 |
+
if scale_ratio > 1:
|
240 |
+
scaled_dims = [scale_ratio * img_dim, img_dim]
|
241 |
+
tile_behavior = "width"
|
242 |
+
elif orig_img_dims[0] < orig_img_dims[1]:
|
243 |
+
scale_ratio = round(orig_img_dims[1] / orig_img_dims[0])
|
244 |
+
if scale_ratio > 1:
|
245 |
+
scaled_dims = [img_dim, scale_ratio * img_dim]
|
246 |
+
tile_behavior = "height"
|
247 |
+
|
248 |
+
resized_image = image.resize(scaled_dims, Image.LANCZOS)
|
249 |
|
250 |
+
if tile_behavior == "width":
|
251 |
+
image_tiles = []
|
252 |
+
for x in range(0, scale_ratio):
|
253 |
+
box = (x * img_dim, 0, (x + 1) * img_dim, img_dim)
|
254 |
+
image_tiles.append(resized_image.crop(box))
|
255 |
|
256 |
+
elif tile_behavior == "height":
|
257 |
+
image_tiles = []
|
258 |
+
for y in range(0, scale_ratio):
|
259 |
+
box = (0, y * img_dim, img_dim, (y + 1) * img_dim)
|
260 |
+
image_tiles.append(resized_image.crop(box))
|
261 |
+
|
262 |
+
else:
|
263 |
+
image_tiles = [resized_image]
|
264 |
+
|
265 |
+
elif st.session_state.vision_mode == "stretched":
|
266 |
+
image_tiles = [image.resize((img_dim, img_dim), Image.LANCZOS)]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
267 |
|
268 |
+
else: # vision_mode == "cropped"
|
269 |
+
if orig_img_dims[0] > orig_img_dims[1]:
|
270 |
+
scale_factor = orig_img_dims[0] / orig_img_dims[1]
|
271 |
+
resized_img_dims = (round(scale_factor * img_dim), img_dim)
|
272 |
resized_img = image.resize(resized_img_dims)
|
273 |
+
elif orig_img_dims[0] < orig_img_dims[1]:
|
274 |
+
scale_factor = orig_img_dims[1] / orig_img_dims[0]
|
275 |
+
resized_img_dims = (img_dim, round(scale_factor * img_dim))
|
276 |
+
else:
|
277 |
+
resized_img_dims = (img_dim, img_dim)
|
278 |
+
|
279 |
+
resized_img = image.resize(resized_img_dims)
|
280 |
+
|
281 |
+
left = round((resized_img_dims[0] - img_dim) / 2)
|
282 |
+
top = round((resized_img_dims[1] - img_dim) / 2)
|
283 |
+
x_right = round(resized_img_dims[0] - img_dim) - left
|
284 |
+
x_bottom = round(resized_img_dims[1] - img_dim) - top
|
285 |
+
right = resized_img_dims[0] - x_right
|
286 |
+
bottom = resized_img_dims[1] - x_bottom
|
287 |
|
288 |
+
# Crop the center of the image
|
289 |
+
image_tiles = [resized_img.crop((left, top, right, bottom))]
|
|
|
|
|
|
|
|
|
290 |
|
291 |
+
image_visualizations = []
|
292 |
+
image_features = []
|
293 |
+
image_similarities = []
|
294 |
|
295 |
+
if st.session_state.active_model == "M-CLIP (multilingual ViT)":
|
296 |
+
text_features = st.session_state.ml_model.forward(
|
297 |
+
st.session_state.search_field_value, st.session_state.ml_tokenizer
|
298 |
+
)
|
299 |
+
|
300 |
+
if st.session_state.device == "cpu":
|
301 |
+
text_features = text_features.float().to(st.session_state.device)
|
302 |
+
else:
|
303 |
+
text_features = text_features.to(st.session_state.device)
|
304 |
|
305 |
+
for altered_image in image_tiles:
|
306 |
+
p_image = (
|
307 |
+
st.session_state.ml_image_preprocess(altered_image)
|
308 |
+
.unsqueeze(0)
|
309 |
+
.to(st.session_state.device)
|
310 |
)
|
311 |
|
312 |
+
vis_t, img_feats, similarity = interpret_vit_overlapped(
|
313 |
+
p_image.type(image_model.dtype),
|
314 |
+
text_features.type(image_model.dtype),
|
315 |
+
image_model.visual,
|
316 |
+
st.session_state.device,
|
317 |
+
img_dim=img_dim,
|
318 |
)
|
319 |
|
320 |
+
image_visualizations.append(vis_t)
|
321 |
+
image_features.append(img_feats)
|
322 |
+
image_similarities.append(similarity.item())
|
323 |
+
|
324 |
+
elif st.session_state.active_model == "J-CLIP (日本語 ViT)":
|
325 |
+
t_text = st.session_state.ja_tokenizer(
|
326 |
+
st.session_state.search_field_value,
|
327 |
+
return_tensors="pt",
|
328 |
+
device=st.session_state.device,
|
329 |
+
)
|
330 |
+
|
331 |
+
text_features = st.session_state.ja_model.get_text_features(**t_text)
|
332 |
+
|
333 |
+
if st.session_state.device == "cpu":
|
334 |
+
text_features = text_features.float().to(st.session_state.device)
|
335 |
+
else:
|
336 |
+
text_features = text_features.to(st.session_state.device)
|
337 |
+
|
338 |
+
for altered_image in image_tiles:
|
339 |
+
p_image = (
|
340 |
+
st.session_state.ja_image_preprocess(altered_image)
|
341 |
+
.unsqueeze(0)
|
342 |
+
.to(st.session_state.device)
|
343 |
)
|
344 |
|
345 |
+
vis_t, img_feats, similarity = interpret_vit_overlapped(
|
346 |
+
p_image.type(image_model.dtype),
|
347 |
+
text_features.type(image_model.dtype),
|
348 |
+
image_model.visual,
|
349 |
+
st.session_state.device,
|
350 |
+
img_dim=img_dim,
|
351 |
)
|
352 |
+
|
353 |
+
image_visualizations.append(vis_t)
|
354 |
+
image_features.append(img_feats)
|
355 |
+
image_similarities.append(similarity.item())
|
356 |
+
|
357 |
+
else: # st.session_state.active_model == Legacy
|
358 |
+
text_features = st.session_state.rn_model(st.session_state.search_field_value)
|
359 |
+
|
360 |
+
if st.session_state.device == "cpu":
|
361 |
+
text_features = text_features.float().to(st.session_state.device)
|
362 |
+
else:
|
363 |
+
text_features = text_features.to(st.session_state.device)
|
364 |
+
|
365 |
+
for altered_image in image_tiles:
|
366 |
+
p_image = (
|
367 |
+
st.session_state.rn_image_preprocess(altered_image)
|
368 |
+
.unsqueeze(0)
|
369 |
+
.to(st.session_state.device)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
370 |
)
|
371 |
|
372 |
+
vis_t = interpret_rn_overlapped(
|
373 |
+
p_image.type(image_model.dtype),
|
374 |
+
text_features.type(image_model.dtype),
|
375 |
+
image_model.visual,
|
376 |
+
GradCAM,
|
377 |
+
st.session_state.device,
|
378 |
+
img_dim=img_dim,
|
379 |
)
|
380 |
|
381 |
+
text_features_norm = text_features.norm(dim=-1, keepdim=True)
|
382 |
+
text_features_new = text_features / text_features_norm
|
383 |
|
384 |
+
image_feats = image_model.encode_image(p_image.type(image_model.dtype))
|
385 |
+
image_feats_norm = image_feats.norm(dim=-1, keepdim=True)
|
386 |
+
image_feats_new = image_feats / image_feats_norm
|
|
|
|
|
|
|
387 |
|
388 |
+
similarity = image_feats_new[0].dot(text_features_new[0])
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
389 |
|
390 |
+
image_visualizations.append(vis_t)
|
391 |
+
image_features.append(p_image)
|
392 |
+
image_similarities.append(similarity.item())
|
393 |
|
394 |
+
transform = ToPILImage()
|
395 |
|
396 |
+
vis_images = [transform(vis_t) for vis_t in image_visualizations]
|
397 |
|
398 |
+
if st.session_state.vision_mode == "cropped":
|
399 |
+
resized_img.paste(vis_images[0], (left, top))
|
400 |
+
vis_images = [resized_img]
|
401 |
|
402 |
+
if orig_img_dims[0] > orig_img_dims[1]:
|
403 |
+
scale_factor = MAX_IMG_WIDTH / orig_img_dims[0]
|
404 |
+
scaled_dims = [MAX_IMG_WIDTH, int(orig_img_dims[1] * scale_factor)]
|
405 |
+
else:
|
406 |
+
scale_factor = MAX_IMG_HEIGHT / orig_img_dims[1]
|
407 |
+
scaled_dims = [int(orig_img_dims[0] * scale_factor), MAX_IMG_HEIGHT]
|
408 |
|
409 |
+
if tile_behavior == "width":
|
410 |
+
vis_image = Image.new("RGB", (len(vis_images) * img_dim, img_dim))
|
411 |
+
for x, v_img in enumerate(vis_images):
|
412 |
+
vis_image.paste(v_img, (x * img_dim, 0))
|
413 |
+
activations_image = vis_image.resize(scaled_dims)
|
414 |
|
415 |
+
elif tile_behavior == "height":
|
416 |
+
vis_image = Image.new("RGB", (img_dim, len(vis_images) * img_dim))
|
417 |
+
for y, v_img in enumerate(vis_images):
|
418 |
+
vis_image.paste(v_img, (0, y * img_dim))
|
419 |
+
activations_image = vis_image.resize(scaled_dims)
|
420 |
|
421 |
+
else:
|
422 |
+
activations_image = vis_images[0].resize(scaled_dims)
|
423 |
|
424 |
+
return activations_image, image_features, np.mean(image_similarities)
|
425 |
+
|
426 |
+
|
427 |
+
def visualize_gradcam(image):
|
428 |
+
if "search_field_value" not in st.session_state:
|
429 |
+
return
|
430 |
+
|
431 |
+
header_cols = st.columns([80, 20], vertical_alignment="bottom")
|
432 |
+
with header_cols[0]:
|
433 |
+
st.title("Image + query details")
|
434 |
+
with header_cols[1]:
|
435 |
+
if st.button("Close"):
|
436 |
+
st.rerun()
|
437 |
+
|
438 |
+
if st.session_state.active_model == "M-CLIP (multilingual ViT)":
|
439 |
+
img_dim = 240
|
440 |
+
image_model = st.session_state.ml_image_model
|
441 |
+
# Sometimes used for token importance viz
|
442 |
+
tokenized_text = st.session_state.ml_tokenizer.tokenize(
|
443 |
+
st.session_state.search_field_value
|
444 |
+
)
|
445 |
+
elif st.session_state.active_model == "Legacy (multilingual ResNet)":
|
446 |
+
img_dim = 288
|
447 |
+
image_model = st.session_state.rn_image_model
|
448 |
+
# Sometimes used for token importance viz
|
449 |
+
tokenized_text = st.session_state.rn_tokenizer.tokenize(
|
450 |
+
st.session_state.search_field_value
|
451 |
+
)
|
452 |
+
else: # J-CLIP
|
453 |
+
img_dim = 224
|
454 |
+
image_model = st.session_state.ja_image_model
|
455 |
+
# Sometimes used for token importance viz
|
456 |
+
tokenized_text = st.session_state.ja_tokenizer.tokenize(
|
457 |
+
st.session_state.search_field_value
|
458 |
)
|
459 |
|
460 |
+
with st.spinner("Calculating..."):
|
461 |
+
# info_text = st.text("Calculating activation regions...")
|
462 |
+
|
463 |
+
activations_image, image_features, similarity_score = get_overlay_vis(
|
464 |
+
image, img_dim, image_model
|
465 |
)
|
466 |
|
467 |
+
st.markdown(
|
468 |
+
f"**Query text:** {st.session_state.search_field_value} | **Approx. image relevance:** {round(similarity_score.item(), 3)}"
|
469 |
+
)
|
470 |
+
|
471 |
+
st.image(activations_image)
|
472 |
+
|
473 |
+
# image_io = BytesIO()
|
474 |
+
# activations_image.save(image_io, "PNG")
|
475 |
+
# dataurl = "data:image/png;base64," + b64encode(image_io.getvalue()).decode(
|
476 |
+
# "ascii"
|
477 |
+
# )
|
478 |
+
|
479 |
+
# st.html(
|
480 |
+
# f"""<div style="display: flex; flex-direction: column; align-items: center;">
|
481 |
+
# <img src="{dataurl}" />
|
482 |
+
# </div>"""
|
483 |
+
# )
|
484 |
+
|
485 |
+
tokenized_text = [
|
486 |
+
tok.replace("▁", "").replace("#", "") for tok in tokenized_text if tok != "▁"
|
487 |
+
]
|
488 |
tokenized_text = [
|
489 |
+
tok
|
490 |
+
for tok in tokenized_text
|
491 |
+
if tok
|
492 |
+
not in ["s", "ed", "a", "the", "an", "ing", "て", "に", "の", "は", "と", "た"]
|
493 |
]
|
494 |
|
495 |
if (
|
|
|
499 |
"Calculate text importance (may take some time)",
|
500 |
)
|
501 |
):
|
502 |
+
scores_per_token = {}
|
|
|
503 |
|
504 |
progress_text = f"Processing {len(tokenized_text)} text tokens"
|
505 |
progress_bar = st.progress(0.0, text=progress_text)
|
|
|
507 |
for t, tok in enumerate(tokenized_text):
|
508 |
token = tok
|
509 |
|
510 |
+
for img_feats in image_features:
|
511 |
+
if st.session_state.active_model == "Legacy (multilingual ResNet)":
|
512 |
+
word_rel = rn_perword_relevance(
|
513 |
+
img_feats,
|
514 |
+
st.session_state.search_field_value,
|
515 |
+
image_model,
|
516 |
+
tokenize,
|
517 |
+
GradCAM,
|
518 |
+
st.session_state.device,
|
519 |
+
token,
|
520 |
+
data_only=True,
|
521 |
+
img_dim=img_dim,
|
522 |
+
)
|
523 |
+
else:
|
524 |
+
word_rel = vit_perword_relevance(
|
525 |
+
img_feats,
|
526 |
+
st.session_state.search_field_value,
|
527 |
+
image_model,
|
528 |
+
tokenize,
|
529 |
+
st.session_state.device,
|
530 |
+
token,
|
531 |
+
img_dim=img_dim,
|
532 |
+
)
|
533 |
+
avg_score = np.mean(word_rel)
|
534 |
+
if avg_score == 0 or np.isnan(avg_score):
|
535 |
+
continue
|
536 |
+
|
537 |
+
if token not in scores_per_token:
|
538 |
+
scores_per_token[token] = [1 / avg_score]
|
539 |
+
else:
|
540 |
+
scores_per_token[token].append(1 / avg_score)
|
541 |
|
542 |
progress_bar.progress(
|
543 |
(t + 1) / len(tokenized_text),
|
|
|
545 |
)
|
546 |
progress_bar.empty()
|
547 |
|
548 |
+
avg_scores_per_token = [
|
549 |
+
np.mean(scores_per_token[tok]) for tok in list(scores_per_token.keys())
|
550 |
+
]
|
551 |
+
|
552 |
+
normed_scores = torch.softmax(torch.tensor(avg_scores_per_token), dim=0)
|
553 |
|
554 |
token_scores = [f"{round(score.item() * 100, 3)}%" for score in normed_scores]
|
555 |
st.session_state.text_table_df = pd.DataFrame(
|
556 |
+
{"token": list(scores_per_token.keys()), "importance": token_scores}
|
557 |
)
|
558 |
|
559 |
st.markdown("**Importance of each text token to relevance score**")
|
560 |
st.table(st.session_state.text_table_df)
|
561 |
|
562 |
|
563 |
+
@st.dialog(" ", width="large")
|
564 |
+
def image_modal(image):
|
565 |
+
visualize_gradcam(image)
|
566 |
|
567 |
|
568 |
+
def vis_known_image(vis_image_id):
|
569 |
+
image_url = st.session_state.images_info.loc[vis_image_id]["image_url"]
|
570 |
+
image_response = requests.get(image_url)
|
571 |
+
image = Image.open(BytesIO(image_response.content), formats=["JPEG", "GIF", "PNG"])
|
572 |
+
image = image.convert("RGB")
|
573 |
+
|
574 |
+
image_modal(image)
|
575 |
+
|
576 |
+
|
577 |
+
def vis_uploaded_image():
|
578 |
+
uploaded_file = st.session_state.uploaded_image
|
579 |
+
if uploaded_file is not None:
|
580 |
+
# To read file as bytes:
|
581 |
+
bytes_data = uploaded_file.getvalue()
|
582 |
+
image = Image.open(BytesIO(bytes_data), formats=["JPEG", "GIF", "PNG"])
|
583 |
+
image = image.convert("RGB")
|
584 |
+
|
585 |
+
image_modal(image)
|
586 |
+
|
587 |
+
|
588 |
+
def format_vision_mode(mode_stub):
|
589 |
+
return mode_stub.capitalize()
|
590 |
|
591 |
|
592 |
st.title("Explore Japanese visual aesthetics with CLIP models")
|
|
|
705 |
use_container_width=True,
|
706 |
)
|
707 |
|
708 |
+
controls = st.columns([25, 25, 20, 35], gap="large", vertical_alignment="center")
|
709 |
with controls[0]:
|
710 |
im_per_pg = st.columns([30, 70], vertical_alignment="center")
|
711 |
with im_per_pg[0]:
|
|
|
715 |
"Images/page:", range(10, 50, 10), label_visibility="collapsed"
|
716 |
)
|
717 |
with controls[1]:
|
|
|
|
|
718 |
im_per_row = st.columns([30, 70], vertical_alignment="center")
|
719 |
with im_per_row[0]:
|
720 |
st.markdown("**Images/row:**")
|
|
|
723 |
"Images/row:", range(1, 6), value=5, label_visibility="collapsed"
|
724 |
)
|
725 |
num_batches = ceil(len(st.session_state.image_ids) / batch_size)
|
726 |
+
with controls[2]:
|
|
|
|
|
727 |
pager = st.columns([40, 60], vertical_alignment="center")
|
728 |
with pager[0]:
|
729 |
st.markdown(f"Page **{st.session_state.current_page}** of **{num_batches}** ")
|
|
|
736 |
label_visibility="collapsed",
|
737 |
key="current_page",
|
738 |
)
|
739 |
+
with controls[3]:
|
740 |
+
st.file_uploader(
|
741 |
+
"Upload an image",
|
742 |
+
type=["jpg", "jpeg", "gif", "png"],
|
743 |
+
key="uploaded_image",
|
744 |
+
label_visibility="collapsed",
|
745 |
+
on_change=vis_uploaded_image,
|
746 |
+
)
|
747 |
|
748 |
|
749 |
if len(st.session_state.search_image_ids) == 0:
|
|
|
780 |
if not RUN_LITE or st.session_state.active_model == "M-CLIP (multilingual ViT)":
|
781 |
st.button(
|
782 |
"Explain this",
|
783 |
+
on_click=vis_known_image,
|
784 |
args=[image_id],
|
785 |
use_container_width=True,
|
786 |
key=image_id,
|
requirements.txt
CHANGED
@@ -1,5 +1,6 @@
|
|
1 |
clip @ git+https://github.com/openai/CLIP.git
|
2 |
ftfy==6.2.0
|
|
|
3 |
multilingual_clip==1.0.10
|
4 |
numpy==1.26
|
5 |
opencv-python==4.10.0.84
|
@@ -7,7 +8,6 @@ pandas==2.1.2
|
|
7 |
pillow==10.1.0
|
8 |
requests==2.31.0
|
9 |
sentencepiece==0.2.0
|
10 |
-
streamlit
|
11 |
torch==2.4.0
|
12 |
torchvision==0.19.0
|
13 |
transformers==4.35.0
|
|
|
1 |
clip @ git+https://github.com/openai/CLIP.git
|
2 |
ftfy==6.2.0
|
3 |
+
matplotlib==3.8.1
|
4 |
multilingual_clip==1.0.10
|
5 |
numpy==1.26
|
6 |
opencv-python==4.10.0.84
|
|
|
8 |
pillow==10.1.0
|
9 |
requests==2.31.0
|
10 |
sentencepiece==0.2.0
|
|
|
11 |
torch==2.4.0
|
12 |
torchvision==0.19.0
|
13 |
transformers==4.35.0
|