CrossOver/configs/preprocess/process_1d.yaml at 2295a966c5708c2530090db6b12b00994ee7f242 · GradientSpaces/CrossOver · GitHub

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
name: "PreProcess1D"

base_dir : "/drive/dumps/multimodal-spaces"
exp_dir  : ""

data:
  sources: []
  process_dir: ${base_dir}/preprocess_feats
  Scannet:
    base_dir       : /drive/datasets/Scannet/
    shape_dir      : /drive/datasets/Shapenet/ShapeNetCore.v2/
    process_dir    : ${data.process_dir}/Scannet/
    processor1D    : Scannet1DProcessor
    processor3D    : Scannet3DProcessor
    mesh_subfix    : _vh_clean_2.labels.ply
    seg_subfix     : _vh_clean_2.0.010000.segs.json
    aggre_subfix   : _vh_clean.aggregation.json

  Scan3R:
    base_dir       : /media/sayan/internal/datasets/Scan3R/
    process_dir    : ${data.process_dir}/Scan3R/
    processor3D    : Scan3R3DProcessor
    processor2D    : Scan3R2DProcessor
    processor1D    : Scan3R1DProcessor
    label_filename : labels.instances.align.annotated.v2.ply
    skip_frames    : 1

  ARKitScenes:
    base_dir       : /media/sayan/internal/datasets/ARKitScenes
    process_dir    : ${data.process_dir}/ARKitScenes/
    processor3D    : ARKitScenes3DProcessor
    processor2D    : ARKitScenes2DProcessor
    processor1D    : ARKitScenes1DProcessor
    skip_frames    : 1
  MultiScan:
    base_dir       : /media/sayan/internal/datasets/MultiScan
    process_dir    : ${data.process_dir}/MultiScan
    processor3D    : MultiScan3DProcessor
    processor2D    : MultiScan2DProcessor
    processor1D    : MultiScan1DProcessor
    skip_frames    : 1

  Structured3D:
    base_dir       : /media/sayan/internal/datasets/Structured3D/
    process_dir    : ${data.process_dir}/Structured3D/scans
    processor3D    : Structured3D_3DProcessor
    processor2D    : Structured3D_2DProcessor
    processor1D    : Structured3D_1DProcessor

  Shapenet:
    base_dir       : /drive/datasets/Shapenet/ShapeNetCore.v2/

modality_info:
  1D  :
    feature_extractor:
      embed_dim : 768
      model_path : 'https://storage.googleapis.com/sfr-vision-language-research/BLIP/models/model_large.pth'

  2D  :
    feature_extractor:
      model     : DinoV2
      ckpt      : dinov2_vitg14
      embed_dim : 1536

  3D  :
    feature_extractor:
      model      : I2PMAE
      ckpt       : /drive/pretrained-models/pointbind_i2pmae.pt
      embed_dim  : 384

    voxel_size : 0.05
    max_points_per_object : 1024
    min_points_per_object : 50

# 'Preprocess' / 'ObjectLevelTrain', 'SceneLevelTrain', 'UnifiedTrain1D', 'UnifiedTrain2D', 'UnifiedTrain3D'
task:
  name       : Preprocess
  Preprocess :
    modality : '1D'
    splits   : ['train', 'val']