Hugging Face's logo

Spaces:
LXT
/
OMG_Seg
Runtime error

App Files Community
4
OMG_Seg
File size: 3,171 Bytes 
b34d1d6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
 
from mmcv import TransformBroadcaster, LoadImageFromFile, RandomResize
from mmdet.datasets.transforms import Resize, RandomFlip, RandomCrop
from mmengine.dataset import DefaultSampler

from seg.datasets.davis import DAVIS
from seg.datasets.pipeliens.frame_copy import AddSemSeg
from seg.datasets.pipeliens.loading import LoadVideoSegAnnotations, ResizeOri
from seg.datasets.pipeliens.formatting import PackVidSegInputs
from seg.datasets.pipeliens.frame_sampling import VideoClipSample
from seg.datasets.samplers.batch_sampler import VideoSegAspectRatioBatchSampler
from seg.evaluation.metrics.vos_metric import VOSMetric

dataset_type = DAVIS
data_root = 'data/DAVIS'

backend_args = None
image_size = (1280, 736)

# dataset settings
train_pipeline = [
    dict(
        type=VideoClipSample,
        num_selected=2,
        interval=2),
    dict(
        type=TransformBroadcaster,
        share_random_params=True,
        transforms=[
            dict(type=LoadImageFromFile, backend_args=backend_args),
            dict(type=LoadVideoSegAnnotations, with_bbox=True, with_label=True, with_mask=True, with_seg=False),
            dict(type=AddSemSeg),
            dict(
                type=RandomResize,
                resize_type=Resize,
                scale=image_size,
                ratio_range=(.9, 1.1),
                keep_ratio=True,
            ),
            dict(
                type=RandomCrop,
                crop_size=image_size,
                crop_type='absolute',
                recompute_bbox=True,
                allow_negative_crop=True),
            dict(type=RandomFlip, prob=0.5),
        ]),
    dict(type=PackVidSegInputs)
]

test_pipeline = [
    dict(
        type=TransformBroadcaster,
        transforms=[
            dict(type=LoadImageFromFile, backend_args=backend_args),
            dict(type=Resize, scale=image_size, keep_ratio=True),
            dict(type=LoadVideoSegAnnotations, with_bbox=True, with_label=True, with_mask=True, with_seg=False),
            # dict(type=ResizeOri),
        ]),
    dict(type=PackVidSegInputs)
]

# dataloader
train_dataloader = dict(
    batch_size=2,
    num_workers=2,
    persistent_workers=True,
    sampler=dict(type=DefaultSampler, shuffle=True),
    batch_sampler=dict(type=VideoSegAspectRatioBatchSampler),
    dataset=dict(
        type=dataset_type,
        data_root=data_root,
        dataset_version='2017',
        ann_file='ImageSets/2017/train.txt',
        data_prefix=dict(img='JPEGImages/Full-Resolution/', ann='Annotations/Full-Resolution/'),
        pipeline=train_pipeline))
val_dataloader = dict(
    batch_size=1,
    num_workers=2,
    persistent_workers=True,
    drop_last=False,
    sampler=dict(type=DefaultSampler, shuffle=False, round_up=False),
    dataset=dict(
        type=dataset_type,
        data_root=data_root,
        dataset_version='2017',
        ann_file='ImageSets/2017/val.txt',
        data_prefix=dict(img='JPEGImages/480p/', ann='Annotations/480p/'),
        test_mode=True,
        pipeline=test_pipeline))
test_dataloader = val_dataloader

val_evaluator = dict(
    type=VOSMetric,
    format_only=True,
)
test_evaluator = val_evaluator