chore: import upstream snapshot with attribution
This commit is contained in:
wehub-resource-sync
@@ -0,0 +1 @@
|
||||
Blechschilder MOORE - CARWASH - Best Hand Job In Town
|
||||
Binary file not shown.
Binary file not shown.
|
After Width: | Height: | Size: 53 KiB |
@@ -0,0 +1 @@
|
||||
{"NSFW": "UNLIKELY", "similarity": 0.33181482553482056, "LICENSE": "?", "caption": "Blechschilder MOORE - CARWASH - Best Hand Job In Town", "url": "https://static.posters.cz/image/350/metallschilder/moore-carwash-best-hand-job-in-town-i19436.jpg", "key": "062690093", "status": "success", "error_message": null, "width": 350, "height": 275, "exif": "{}", "original_width": 350, "original_height": 275}
|
||||
@@ -0,0 +1,5 @@
|
||||
in 343,454,384,439,394,489,354,502 0.9964783191680908
|
||||
town! 387,431,492,396,504,450,399,485 0.9821552038192749
|
||||
Best 2,206,96,173,112,243,19,277 0.995647132396698
|
||||
hand 102,173,193,139,207,201,115,234 0.9959321618080139
|
||||
job 194,143,254,117,272,184,212,210 0.9975775480270386
|
||||
@@ -0,0 +1 @@
|
||||
Weekender: Win 40K Shadowspear & GW TOP TRUMP?
|
||||
Binary file not shown.
Binary file not shown.
|
After Width: | Height: | Size: 60 KiB |
@@ -0,0 +1 @@
|
||||
{"NSFW": "UNLIKELY", "similarity": 0.30257895588874817, "LICENSE": "?", "caption": "Weekender: Win 40K Shadowspear & GW TOP TRUMP?", "url": "https://images.beastsofwar.com/2019/02/Weekender-260-Front-v2-Cover-Image-1024x576.jpg", "key": "062692210", "status": "success", "error_message": null, "width": 1024, "height": 576, "exif": "{}", "original_width": 1024, "original_height": 576}
|
||||
@@ -0,0 +1,7 @@
|
||||
an 439,418,466,379,497,448,470,487 0.9474218487739563
|
||||
DOWNSPEAR 25,440,185,382,190,421,30,480 0.5379968881607056
|
||||
WIN 15,354,91,327,101,414,25,441 0.9966829419136047
|
||||
Love 185,336,192,323,211,352,204,367 0.414926141500473
|
||||
40K 99,312,183,294,189,389,105,406 0.971768319606781
|
||||
T 184,264,218,214,295,379,262,429 0.11518792808055878
|
||||
CHAT 14,20,120,20,120,102,14,102 0.9378468990325928
|
||||
@@ -0,0 +1 @@
|
||||
"Little Joe - ""A La Guerra Me Llevan"""
|
||||
Binary file not shown.
Binary file not shown.
|
After Width: | Height: | Size: 30 KiB |
@@ -0,0 +1 @@
|
||||
{"NSFW": "UNLIKELY", "similarity": 0.3536737263202667, "LICENSE": "?", "caption": "\"Little Joe - \"\"A La Guerra Me Llevan\"\"\"", "url": "https://i.ytimg.com/vi/8bx7Wx59Oxc/hqdefault.jpg", "key": "062692530", "status": "success", "error_message": null, "width": 480, "height": 360, "exif": "{}", "original_width": 480, "original_height": 360}
|
||||
@@ -0,0 +1,3 @@
|
||||
FAMILIA 185,376,186,277,437,283,436,382 0.9923758506774902
|
||||
LITTLE 37,231,40,119,243,129,241,240 0.99787437915802
|
||||
JOE 257,236,261,105,393,113,389,244 0.996442437171936
|
||||
@@ -0,0 +1 @@
|
||||
VH-EZT at Busselton Airport Tail shot
|
||||
Binary file not shown.
Binary file not shown.
|
After Width: | Height: | Size: 20 KiB |
@@ -0,0 +1 @@
|
||||
{"NSFW": "UNLIKELY", "similarity": 0.3146217465400696, "LICENSE": "?", "caption": "VH-EZT at Busselton Airport Tail shot", "url": "http://uniflying.org.au/wp-content/uploads/EZT-STYLE-2-150x150.jpg", "key": "271975131", "status": "success", "error_message": null, "width": 512, "height": 512, "original_width": 150, "original_height": 150, "exif": "{}", "sha256": "b32fe5c60a42f18fd496a461c948683cd1f52e89033a61660795a6d2cd3b93de"}
|
||||
@@ -0,0 +1 @@
|
||||
VH-EZT 143,133,501,82,532,301,175,352 0.21680155396461487
|
||||
@@ -0,0 +1 @@
|
||||
Strangers to These Shores : Race and Ethnic Relations in the United States, Parrillo, Vincent N., 0205585574
|
||||
Binary file not shown.
Binary file not shown.
|
After Width: | Height: | Size: 32 KiB |
@@ -0,0 +1 @@
|
||||
{"NSFW": "UNLIKELY", "similarity": 0.30967631936073303, "LICENSE": "?", "caption": "Strangers to These Shores : Race and Ethnic Relations in the United States, Parrillo, Vincent N., 0205585574", "url": "https://dcqohhuh7tdzn.cloudfront.net/5571/9780205585571_medium.jpg", "key": "271975251", "status": "success", "error_message": null, "width": 512, "height": 512, "original_width": 127, "original_height": 160, "exif": "{}", "sha256": "86d27d0284bfd0f6d01ce332bd157e01d43265d48e9a20646321c6ad7c070b35"}
|
||||
@@ -0,0 +1 @@
|
||||
SHORES 148,89,361,83,362,140,150,146 0.9963237643241882
|
||||
@@ -0,0 +1 @@
|
||||
Marvel MV15598 Manta Polar, Multicolor, 150x100cm
|
||||
Binary file not shown.
Binary file not shown.
|
After Width: | Height: | Size: 36 KiB |
@@ -0,0 +1 @@
|
||||
{"NSFW": "UNLIKELY", "similarity": 0.31281691789627075, "LICENSE": "?", "caption": "Marvel MV15598 Manta Polar, Multicolor, 150x100cm", "url": "https://m.media-amazon.com/images/I/51vgtDUtImL._AC_US200_.jpg", "key": "271978467", "status": "success", "error_message": null, "width": 512, "height": 512, "original_width": 200, "original_height": 200, "exif": "{}", "sha256": "039cf017effd4aeac9b2b8511f4a62ef3afc6b7af0bbee42eb0b5fb03b3896cc"}
|
||||
@@ -0,0 +1,5 @@
|
||||
unr 357,447,362,421,414,433,408,459 0.28840482234954834
|
||||
PIDERMA 71,403,409,320,432,415,94,498 0.6835125088691711
|
||||
MARVEL 211,331,301,320,307,363,217,374 0.9939236640930176
|
||||
GREAT 361,74,417,74,417,99,361,99 0.9460398554801941
|
||||
COMES 312,54,324,39,373,78,361,93 0.9603791236877441
|
||||
@@ -0,0 +1,32 @@
|
||||
import json
|
||||
import os
|
||||
import traceback
|
||||
from tqdm import tqdm
|
||||
from multiprocessing import Pool
|
||||
|
||||
ROOT_FROM = 'XXX' # the path of laion-ocr-zip
|
||||
ROOT_TO = 'XXX' # the path for saving dataset
|
||||
MULTIPROCESSING_NUM = 64
|
||||
DOWNLOAD_IMAGES = False # whether to download images from urls
|
||||
|
||||
def unzip_file(idx):
|
||||
if not os.path.exists(f'{ROOT_FROM}/{idx}.zip') or os.path.exists(f'{ROOT_TO}/{idx}'):
|
||||
return
|
||||
cmd = f'unzip -q {ROOT_FROM}/{idx}.zip -d {ROOT_TO}'
|
||||
os.system(cmd)
|
||||
|
||||
|
||||
def multiprocess_unzip_file(idxs):
|
||||
os.makedirs(ROOT_TO, exist_ok=True)
|
||||
|
||||
with Pool(processes=MULTIPROCESSING_NUM) as p:
|
||||
with tqdm(total=len(idxs), desc='total') as pbar:
|
||||
for i, _ in enumerate(p.imap_unordered(unzip_file, idxs)):
|
||||
pbar.update()
|
||||
print("multiprocess_unzip_file done!")
|
||||
|
||||
if __name__ == '__main__':
|
||||
files = os.listdir(ROOT_FROM)
|
||||
idxs = [str(idx[:-4]).zfill(5) for idx in files]
|
||||
multiprocess_unzip_file(idxs)
|
||||
print("Finished!")
|
||||
File diff suppressed because one or more lines are too long
Reference in New Issue
Block a user