File size: 5,578 Bytes
e4bd7f9
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
import os
import logging
import warnings

from bubogpt.common.registry import registry
from bubogpt.datasets.builders.image_base_dataset_builder import ImageBaseDatasetBuilder
from bubogpt.datasets.datasets.image_caption.laion_dataset import LaionDataset
from bubogpt.datasets.datasets.image_caption.cc_sbu_dataset import CCSBUDataset, \
    CCSBUAlignDatasetImageImageCaptionDataset, CCDataset
from bubogpt.datasets.datasets.image_caption.llava_dataset import LlavaInstruct150Dataset

@registry.register_builder("cc_sbu")
class CCSBUBuilderImage(ImageBaseDatasetBuilder):
    train_dataset_cls = CCSBUDataset

    DATASET_CONFIG_DICT = {"default": "configs/datasets/cc_sbu/defaults.yaml"}

    def _download_ann(self):
        pass

    def _download_vis(self):
        pass

    def build(self):
        self.build_processors()

        build_info = self.config.build_info

        datasets = dict()
        split = "train"

        # create datasets
        # [NOTE] return inner_datasets (wds.DataPipeline)
        dataset_cls = self.train_dataset_cls
        datasets[split] = dataset_cls(
            vision_processor=self.vis_processors[split],
            text_processor=self.text_processors[split],
            location=build_info.storage,
        ).inner_dataset

        return datasets


@registry.register_builder("laion")
class LaionBuilderImage(ImageBaseDatasetBuilder):
    train_dataset_cls = LaionDataset

    DATASET_CONFIG_DICT = {"default": "configs/datasets/laion/defaults.yaml"}

    def _download_ann(self):
        pass

    def _download_vis(self):
        pass

    def build(self):
        self.build_processors()

        build_info = self.config.build_info

        datasets = dict()
        split = "train"

        # create datasets
        # [NOTE] return inner_datasets (wds.DataPipeline)
        dataset_cls = self.train_dataset_cls
        datasets[split] = dataset_cls(
            vision_processor=self.vis_processors[split],
            text_processor=self.text_processors[split],
            location=build_info.storage,
        ).inner_dataset

        return datasets


@registry.register_builder("cc_sbu_align")
class CCSBUAlignBuilderImage(ImageBaseDatasetBuilder):
    train_dataset_cls = CCSBUAlignDatasetImageImageCaptionDataset

    DATASET_CONFIG_DICT = {
        "default": "configs/datasets/cc_sbu/align.yaml",
    }

    def build_datasets(self):
        # at this point, all the annotations and image/videos should be all downloaded to the specified locations.
        logging.info("Building datasets...")
        self.build_processors()

        build_info = self.config.build_info
        storage_path = build_info.storage

        datasets = dict()

        if not os.path.exists(storage_path):
            warnings.warn("storage path {} does not exist.".format(storage_path))

        # create datasets
        dataset_cls = self.train_dataset_cls
        datasets['train'] = dataset_cls(
            vision_processor=self.vis_processors["train"],
            text_processor=self.text_processors["train"],
            ann_paths=[os.path.join(storage_path, 'filter_cap.json')],
            vis_root=os.path.join(storage_path, 'image'),
        )

        return datasets


@registry.register_builder("cc12m")
class CC12MBuilder(ImageBaseDatasetBuilder):
    train_dataset_cls = CCDataset

    DATASET_CONFIG_DICT = {"default": "configs/datasets/cc12m/defaults.yaml"}

    def _download_ann(self):
        pass

    def _download_vis(self):
        pass

    def build(self):
        self.build_processors()

        build_info = self.config.build_info

        datasets = dict()
        split = "train"

        # create datasets
        # [NOTE] return inner_datasets (wds.DataPipeline)
        dataset_cls = self.train_dataset_cls
        datasets[split] = dataset_cls(
            vis_processor=self.vis_processors[split],
            text_processor=self.text_processors[split],
            location=build_info.storage,
        ).inner_dataset

        return datasets
    

@registry.register_builder("llava_instruct150")
class LlavaInstruct150Builder(ImageBaseDatasetBuilder):
    train_dataset_cls = LlavaInstruct150Dataset

    DATASET_CONFIG_DICT = {"default": None}

    def _download_ann(self):
        pass

    def _download_vis(self):
        pass

    
    def build(self):
        self.build_processors()

        datasets = dict()
        split = "train"
        dataset_cls = self.train_dataset_cls
        datasets[split] = dataset_cls(
            vis_processor=self.vis_processors[split],
            text_processor=self.text_processors[split],
            vis_root="/path/to/dataset/COCO_2014",
            ann_paths=[os.path.join("/path/to/dataset/llava/annotations", subset + '.json')
                       for subset in ["complex_reasoning_77k", "conversation_58k", "detail_23k"]],
        )
        return datasets


# from bubogpt.datasets.builders.image_text_pair_builder import LlavaInstruct150Builder

if __name__ == "__main__":
    from omegaconf import OmegaConf
    from itertools import islice

    data_cfg = OmegaConf.create({
        "vis_processor": {"train": {"name": "imagebind_vision_train", "image_size": 224}},
        "text_processor": {"train": {"name": "imagebind_caption"}},
        "data_type": "image",
        })

    builder = LlavaInstruct150Builder(data_cfg)

    datasets = builder.build_datasets()

    datasets["train"].check_existence()

    for sample in islice(datasets["train"], 10):
        print(sample["vision"].shape, sample["prompt"], sample["text_input"])