Hugging Face's logo

Spaces:
mshukor
/
UnIVAL
Running

App Files Community
UnIVAL
File size: 3,882 Bytes 
26fd00c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
 
from utils import get_tsv_data_from_jsons, create_imagenet_txt_files
import csv
from io import StringIO
from tqdm import tqdm 

        
# with image conversion
# datasets = ['/data/mshukor/data/our_albef_data/json_pretrain/vg_albef.json',
#             '/data/mshukor/data/our_albef_data/json_pretrain/sbu.json',
#             ]

# output_paths = ['/data/mshukor/data/ofa/pretrain_ours/vg_albef.tsv', 
#                '/data/mshukor/data/ofa/pretrain_ours/sbu.tsv',
#                ]

# task_types = ['caption', 
#              'caption']

# start_id = 566747
# for data, task_type, output_path in zip(datasets, task_types, output_paths):

    

#     tsvs = get_tsv_data_from_jsons([data], start_id, [task_type])

#     start_id = tsvs[-1][0] + 1
    
#     print("save tsv to:", output_path)

#     with open(output_path, 'w', newline='') as f_output:
#         csv_output = csv.writer(f_output, delimiter='\t')
#         for t in tqdm(tsvs):
#             csv_output.writerow(t)

########################################################
# without image conversion
    
# datasets = ['/data/mshukor/data/our_albef_data/json_pretrain/coco_karp.json',
#             '/data/mshukor/data/our_albef_data/json_pretrain/vg_albef.json',
#             '/data/mshukor/data/our_albef_data/json_pretrain/sbu.json',
#             '/data/mshukor/data/our_albef_data/json_pretrain/cc3m.json']

# start_id = 0
# task_types = ['caption',
#              'caption',
#              'caption',
#              'caption']

# tsvs = get_tsv_data_from_jsons(datasets, start_id, task_types, convert_images=False)

# output_path = '/data/mshukor/data/ofa/pretrain_ours/vision_language_4m.tsv'

# with open(output_path, 'w', newline='') as f_output:
#     csv_output = csv.writer(f_output, delimiter='\t')

#     for t in tqdm(tsvs):
#         csv_output.writerow(t)
        
        
        
########################################################

        
# datasets = [
#             '/data/mshukor/data/our_albef_data/json_pretrain/coco_karp.json',
#             '/data/mshukor/data/our_albef_data/json_pretrain/vg_albef.json',
#             '/data/mshukor/data/our_albef_data/json_pretrain/sbu.json',
#             '/data/mshukor/data/our_albef_data/json_pretrain/cc3m.json', 
    
#             ['/data/mshukor/data/refcoco/refcoco+/refs(unc).p', '/data/mshukor/data/refcoco/refcoco+/instances.json'],
            
#             '/data/mshukor/data/our_albef_data/data/vqa_train.json',
# ]

# start_id = 0
# task_types = ['caption',
#              'caption',
#              'caption',
#              'caption',
#              'visual_grounding',
#              'qa',]

# tsvs = get_tsv_data_from_jsons(datasets, start_id, task_types, convert_images=False)


# output_path = '/data/mshukor/data/ofa/pretrain_ours/vision_language_mini.tsv'

# with open(output_path, 'w', newline='') as f_output:
#     csv_output = csv.writer(f_output, delimiter='\t')

#     for t in tqdm(tsvs):
#         csv_output.writerow(t)



#### imagenet

path_data = '/data/mshukor/data/imagenet/val'
output_path = '/data/mshukor/data/ofa/pretrain_ours/imagenet_val.txt'


create_imagenet_txt_files(path_data, output_path)



####### object detection



        
from preprocess.utils import get_tsv_data_from_jsons

datasets = [
    ['coco', '/data/mshukor/data/coco/annotations/instances_train2014.json'],
    ['vg', '/data/mshukor/data/visual_genome/annotations/objects.json', '/data/mshukor/data/visual_genome/images'],
]

start_id = 0
task_types = ['detection', 
              'detection',]



tsvs = get_tsv_data_from_jsons(datasets, start_id, task_types, convert_images=False)


output_path = '/data/mshukor/data/ofa/pretrain_ours/detection_mini.tsv'

with open(output_path, 'w', newline='') as f_output:
    csv_output = csv.writer(f_output, delimiter='\t')

    for t in tqdm(tsvs):
        csv_output.writerow(t)