File size: 1,808 Bytes
3ef1661 |
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 |
_base_=['../_base_/losses/all_losses.py',
'../_base_/models/encoder_decoder/dino_vit_large_reg.dpt_raft.py',
'../_base_/datasets/scannet.py',
'../_base_/datasets/scannet_all.py',
#'../_base_/datasets/_data_base_.py',
'../_base_/default_runtime.py',
'../_base_/schedules/schedule_1m.py'
]
import numpy as np
model = dict(
decode_head=dict(
type='RAFTDepthNormalDPT5',
iters=8,
n_downsample=2,
detach=False,
)
)
# model settings
find_unused_parameters = True
# data configs, some similar data are merged together
data_array = [
# group 1
[
#dict(ScanNet='ScanNet_dataset'),
dict(ScanNetAll='ScanNetAll_dataset')
],
]
data_basic=dict(
canonical_space = dict(
# img_size=(540, 960),
focal_length=1000.0,
),
depth_range=(0,1),
depth_normalize=(0.1, 200),
crop_size = (1120, 2016),
clip_depth_range=(0.1, 200),
vit_size=(616,1064),
)
test_metrics = ['abs_rel', 'rmse', 'silog', 'delta1', 'delta2', 'delta3', 'rmse_log', 'log10', 'normal_mean', 'normal_rmse', 'normal_median', 'normal_a3', 'normal_a4', 'normal_a5']
ScanNetAll_dataset=dict(
#ScanNet_dataset=dict(
data = dict(
test=dict(
pipeline=[dict(type='BGR2RGB'),
dict(type='LabelScaleCononical'),
dict(type='ResizeKeepRatio',
resize_size=(616, 1064), #(544, 992), #(480, 1216), #(480, 640), #
ignore_label=-1,
padding=[0,0,0]),
dict(type='ToTensor'),
dict(type='Normalize', mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375]),
],
sample_ratio = 1.0,
sample_size = 500,
),
)) |