Hugging Face's logo

Spaces:
Hancy
/
LiDAR-Diffusion
Running

App Files Community
LiDAR-Diffusion
File size: 2,613 Bytes 
851751e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3274cc5
851751e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
 
model:
  base_learning_rate: 1.0e-06
  target: lidm.models.diffusion.ddpm.LatentDiffusion
  params:
    linear_start: 0.0015
    linear_end: 0.0205
    num_timesteps_cond: 1
    log_every_t: 100
    timesteps: 1000
    image_size: [16, 128]
    channels: 8
    monitor: val/loss_simple_ema
    first_stage_key: image
    cond_stage_key: segmentation
    concat_mode: true
    cond_stage_trainable: true
    verbose: false
    unet_config:
      target: lidm.modules.diffusion.openaimodel.UNetModel
      params:
        image_size: [16, 128]
        in_channels: 16
        out_channels: 8
        model_channels: 256
        attention_resolutions: [4, 2, 1]
        num_res_blocks: 2
        channel_mult: [1, 2, 4]
        num_head_channels: 32
        lib_name: lidm
    first_stage_config:
      target: lidm.models.autoencoder.VQModelInterface
      params:
        embed_dim: 8
        n_embed: 16384
        lib_name: lidm
        use_mask: False  # False
        ckpt_path: models/first_stage_models/kitti/f_c2_p4_wo_logscale/model.ckpt
        ddconfig:
          double_z: false
          z_channels: 8
          in_channels: 1
          out_ch: 1
          ch: 64
          ch_mult: [1,2,2,4]
          strides: [[1,2],[2,2],[2,2]]
          num_res_blocks: 2
          attn_levels: []
          dropout: 0.0
        lossconfig:
          target: torch.nn.Identity
    cond_stage_config:
      target: lidm.modules.encoders.modules.SpatialRescaler
      params:
        strides: [[1,2],[2,2],[2,2]]
        in_channels: 20
        out_channels: 8

data:
  target: main.DataModuleFromConfig
  params:
    batch_size: 16
    num_workers: 8
    wrap: true
    dataset:
      size: [64, 1024]
      fov: [ 3,-25 ]
      depth_range: [ 1.0,56.0 ]
      depth_scale: 56  # np.log2(depth_max + 1)
      log_scale: false
      x_range: [ -50.0, 50.0 ]
      y_range: [ -50.0, 50.0 ]
      z_range: [ -3.0, 1.0 ]
      resolution: 1
      num_channels: 1
      num_cats: 10
      num_views: 2
      num_sem_cats: 19
      filtered_map_cats: [ ]
    aug:
      flip: true
      rotate: false
      keypoint_drop: false
      keypoint_drop_range: [ 5,20 ]
      randaug: false
    train:
      target: lidm.data.kitti.SemanticKITTITrain
      params:
        condition_key: segmentation
    validation:
      target: lidm.data.kitti.SemanticKITTIValidation
      params:
        condition_key: segmentation


lightning:
  callbacks:
    image_logger:
      target: main.ImageLogger
      params:
        batch_frequency: 5000
        max_images: 8
        increase_log_steps: False

  trainer:
    benchmark: true