bev-project/configs/nuscenes/default.yaml

dataset_type: NuScenesDataset
dataset_root: data/nuscenes/
gt_paste_stop_epoch: -1
reduce_beams: 32
load_dim: 5
use_dim: 5
load_augmented: null

point_cloud_range: [-51.2, -51.2, -5.0, 51.2, 51.2, 3.0]
voxel_size: [0.1, 0.1, 0.2]
image_size: [256, 704]

augment2d:
  resize: [[0.38, 0.55], [0.48, 0.48]]
  rotate: [-5.4, 5.4]
  gridmask:
    prob: 0.0
    fixed_prob: true

augment3d:
  scale: [0.9, 1.1]
  rotate: [-0.78539816, 0.78539816]
  translate: 0.5

object_classes:
  - car
  - truck
  - construction_vehicle
  - bus
  - trailer
  - barrier
  - motorcycle
  - bicycle
  - pedestrian
  - traffic_cone

map_classes:
  - drivable_area
  # - drivable_area*
  - ped_crossing
  - walkway
  - stop_line
  - carpark_area
  # - road_divider
  # - lane_divider
  - divider

input_modality:
  use_lidar: true
  use_camera: true
  use_radar: false
  use_map: false
  use_external: false

train_pipeline:
  -
    type: LoadMultiViewImageFromFiles
    to_float32: true
  -
    type: LoadPointsFromFile
    coord_type: LIDAR
    load_dim: ${load_dim}
    use_dim: ${use_dim}
    reduce_beams: ${reduce_beams}
    load_augmented: ${load_augmented}
  -
    type: LoadPointsFromMultiSweeps
    sweeps_num: 9
    load_dim: ${load_dim}
    use_dim: ${use_dim}
    reduce_beams: ${reduce_beams}
    pad_empty_sweeps: true
    remove_close: true
    load_augmented: ${load_augmented}
  -
    type: LoadAnnotations3D
    with_bbox_3d: true
    with_label_3d: true
    with_attr_label: False
  -
    type: ObjectPaste
    stop_epoch: ${gt_paste_stop_epoch}
    db_sampler:
      dataset_root: ${dataset_root}
      info_path: ${dataset_root + "nuscenes_dbinfos_train.pkl"}
      rate: 1.0
      prepare: 
        filter_by_difficulty: [-1]
        filter_by_min_points:
          car: 5
          truck: 5
          bus: 5
          trailer: 5
          construction_vehicle: 5
          traffic_cone: 5
          barrier: 5
          motorcycle: 5
          bicycle: 5
          pedestrian: 5
      classes: ${object_classes}
      sample_groups:
        car: 2
        truck: 3
        construction_vehicle: 7
        bus: 4
        trailer: 6
        barrier: 2
        motorcycle: 6
        bicycle: 6
        pedestrian: 2
        traffic_cone: 2
      points_loader:
        type: LoadPointsFromFile
        coord_type: LIDAR
        load_dim: ${load_dim}
        use_dim: ${use_dim}
        reduce_beams: ${reduce_beams}
  -
    type: ImageAug3D
    final_dim: ${image_size}
    resize_lim: ${augment2d.resize[0]}
    bot_pct_lim: [0.0, 0.0]
    rot_lim: ${augment2d.rotate}
    rand_flip: true
    is_train: true
  -
    type: GlobalRotScaleTrans
    resize_lim: ${augment3d.scale}
    rot_lim: ${augment3d.rotate}
    trans_lim: ${augment3d.translate}
    is_train: true
  -
    type: LoadBEVSegmentation
    dataset_root: ${dataset_root}
    xbound: [-50.0, 50.0, 0.5]
    ybound: [-50.0, 50.0, 0.5]
    classes: ${map_classes}
  -
    type: RandomFlip3D
  -
    type: PointsRangeFilter
    point_cloud_range: ${point_cloud_range}
  -
    type: ObjectRangeFilter
    point_cloud_range: ${point_cloud_range}
  -
    type: ObjectNameFilter
    classes: ${object_classes}
  -
    type: ImageNormalize
    mean: [0.485, 0.456, 0.406]
    std: [0.229, 0.224, 0.225]
  - 
    type: GridMask
    use_h: true
    use_w: true
    max_epoch: ${max_epochs}
    rotate: 1
    offset: false
    ratio: 0.5
    mode: 1
    prob: ${augment2d.gridmask.prob}
    fixed_prob: ${augment2d.gridmask.fixed_prob}
  -
    type: PointShuffle
  -
    type: DefaultFormatBundle3D
    classes: ${object_classes}
  -
    type: Collect3D
    keys:
      - img
      - points
      - gt_bboxes_3d
      - gt_labels_3d
      - gt_masks_bev
    meta_keys:
      - camera_intrinsics
      - camera2ego
      - lidar2ego
      - lidar2camera
      - camera2lidar
      - lidar2image
      - img_aug_matrix
      - lidar_aug_matrix
  -
    type: GTDepth
    keyframe_only: true 

test_pipeline:
  -
    type: LoadMultiViewImageFromFiles
    to_float32: true
  -
    type: LoadPointsFromFile
    coord_type: LIDAR
    load_dim: ${load_dim}
    use_dim: ${use_dim}
    reduce_beams: ${reduce_beams}
    load_augmented: ${load_augmented}
  -
    type: LoadPointsFromMultiSweeps
    sweeps_num: 9
    load_dim: ${load_dim}
    use_dim: ${use_dim}
    reduce_beams: ${reduce_beams}
    pad_empty_sweeps: true
    remove_close: true
    load_augmented: ${load_augmented}
  -
    type: LoadAnnotations3D
    with_bbox_3d: true
    with_label_3d: true
    with_attr_label: False
  -
    type: ImageAug3D
    final_dim: ${image_size}
    resize_lim: ${augment2d.resize[1]}
    bot_pct_lim: [0.0, 0.0]
    rot_lim: [0.0, 0.0]
    rand_flip: false
    is_train: false
  -
    type: GlobalRotScaleTrans
    resize_lim: [1.0, 1.0]
    rot_lim: [0.0, 0.0]
    trans_lim: 0.0
    is_train: false
  -
    type: LoadBEVSegmentation
    dataset_root: ${dataset_root}
    xbound: [-50.0, 50.0, 0.5]
    ybound: [-50.0, 50.0, 0.5]
    classes: ${map_classes}
  -
    type: PointsRangeFilter
    point_cloud_range: ${point_cloud_range}
  -
    type: ImageNormalize
    mean: [0.485, 0.456, 0.406]
    std: [0.229, 0.224, 0.225]
  -
    type: DefaultFormatBundle3D
    classes: ${object_classes}
  -
    type: Collect3D
    keys:
      - img
      - points
      - gt_bboxes_3d
      - gt_labels_3d
      - gt_masks_bev
    meta_keys:
      - camera_intrinsics
      - camera2ego
      - lidar2ego
      - lidar2camera
      - camera2lidar
      - lidar2image
      - img_aug_matrix
      - lidar_aug_matrix
  -
    type: GTDepth
    keyframe_only: true 

data:
  samples_per_gpu: 4
  workers_per_gpu: 4
  train:
    type: CBGSDataset
    dataset:
      type: ${dataset_type}
      dataset_root: ${dataset_root}
      ann_file: ${dataset_root + "nuscenes_infos_train.pkl"}
      pipeline: ${train_pipeline}
      object_classes: ${object_classes}
      map_classes: ${map_classes}
      modality: ${input_modality}
      test_mode: false
      use_valid_flag: true
      box_type_3d: LiDAR
  val:
    type: ${dataset_type}
    dataset_root: ${dataset_root}
    ann_file: ${dataset_root + "nuscenes_infos_val.pkl"}
    pipeline: ${test_pipeline}
    object_classes: ${object_classes}
    map_classes: ${map_classes}
    modality: ${input_modality}
    test_mode: false
    box_type_3d: LiDAR
  test:
    type: ${dataset_type}
    dataset_root: ${dataset_root}
    ann_file: ${dataset_root + "nuscenes_infos_val.pkl"}
    pipeline: ${test_pipeline}
    object_classes: ${object_classes}
    map_classes: ${map_classes}
    modality: ${input_modality}
    test_mode: true
    box_type_3d: LiDAR

evaluation:
  interval: 1
  pipeline: ${test_pipeline}