dynamic_visual_sgraphs/config/system_params.yaml at master · snt-arg/dynamic_visual_sgraphs · 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
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
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
################################ General Parameters #############################################
general:
  # Specify the mode of operation for the segmentation pipeline
  # 0: Semantic Segmentation runs in conjunction with Geometric Segmentation
  # 1: Semantic Segmentation runs independently
  # 2: Geometric Segmentation runs independently
  # run with 2 if semantic segmentation is not running - this clears the pointclouds
  mode_of_operation: 1

  # The path of the high level semantic information of the environment
  # [TODO] - the path looks ugly, possible workaround?
  env_database: ""

#################################### Marker Parameters ###########################################
markers:
  # How much the framework should trust the markers (0.1 default, higher for mono setups)
  impact: 0.1

################################## Optimization Parameters #######################################
optimization:
  marginalize_planes: false

  # plane to map point factor (needs marginalize_planes to be false)
  plane_map_point:
    enabled: false
    information_gain: 1.0
  # plane equation to keyframe factor
  plane_kf:
    enabled: false
    information_gain: 0.5
  # point to plane factor
  plane_point:
    enabled: true
    information_gain: 0.3

################################ Refining map points using semantics Parameters ##################
refine_map_points:
  enabled: true
  max_distance_for_delete: 0.1
  octree:
    resolution: 0.03
    search_radius: 0.03
    min_neighbors: 2

############################### Plane based covisibility graph Parameters #########################
plane_based_covisibility:
  enabled: false
  max_keyframes: 80 # the maximum number of keyframes to be considered for the covisibility graph
  score_per_plane: 80 # the score each semantic plane contributes to the covisibility graph

################################ Common Segmentation Parameters ##################################
seg:
  # Number of points (after filtering) needed to detect geometric objects (e.g., planes)
  # Also, the number of points needed to detect a plane in the pointcloud
  pointclouds_thresh: 250

  # Plane association
  plane_association:
    ominus_thresh: 0.20
    distance_thresh: 0.15 # meters
    centroid_thresh: 2.0 # meters
    cluster_separation:
      enabled: true
      tolerance: 2.0 # euclidean distance for cutting planes
      downsample:
        leaf_size: 0.12
        min_points_per_voxel: 20

  # The maximum distance for point to be considered part of a plane
  plane_point_dist_thresh: 0.02

  ransac:
    # The maximum number of planes to be detected in one pointcloud
    max_planes: 2

    # The maximum distance for a point to be considered an inlier of the plane model
    distance_thresh: 0.02

    # The maximum number of iterations for the RANSAC algorithm
    max_iterations: 800

############################ Geometric Segmentation Parameters ####################################
geo_seg:
  pointcloud:
    # pointcloud downsampling parmeters
    downsample:
      leaf_size: 0.08
      min_points_per_voxel: 10
    # Pointcloud outlier removal
    outlier_removal:
      mean_threshold: 50
      std_threshold: 1.0

############################## Semantic Segmentation Parameters ####################################
sem_seg:
  pointcloud:
    # pointcloud downsampling parmeters
    downsample:
      leaf_size: 0.04
      min_points_per_voxel: 10
    # Pointcloud outlier removal
    outlier_removal:
      mean_threshold: 50
      std_threshold: 1.0

  # The threshold for semantic segmentation probability validation (rejects probability below this)
  prob_thresh: 0.5

  # The threshold for semantic segmentation confidence validation (rejects confidence below this)
  conf_thresh: 0.1

  # The maximum height above the main ground to be a valid stepped ground plane (in meters)
  max_step_elevation: 0.5

  # The maximum tilt heuristic for a wall/ground plane (checks respective component of plane equation)
  max_tilt_wall: 0.2
  max_tilt_ground: 0.2

  # The amount of votes (votes are weighted votes) for a plane to be finally classified with a semantic label
  min_votes: 1.8

  # Whether to reassociate planes with the semantic labels - after optimization, some planes may get close
  reassociate:
    enabled: true
    association_thresh: 0.1

  ############################## Room Segmentation Parameters ####################################
room_seg:
  # The method to use for room segmentation
  # 0: Segment geometrically based on closest walls
  # 1: Segment based on free-space clustering (skeleton voxblox)
  # 2: Segment based on GNN (legacy)
  # 3: Segment based on GNN (new)
  method: 1

  # The minimum space between two given walls of a corridor/room (in meters) to be valid
  min_wall_distance_thresh: 1.0

  # The maximum threshold for the dot product of the plane normals to be considered facing each other
  plane_facing_dot_thresh: -0.9

  # Threshold for perpendicularity of two walls of a room (in degrees)
  perpendicularity_thresh: 10.0

  # Threshold for parallelism of two walls of a room (in degrees)
  parallelism_thresh: 10.0

  # The maximum distance between two room centroids to be considered the same room (in meters)
  center_distance_thresh: 2.0

  # [deprecated] Parameters for the geometrically-based closest walls room segmentation
  geo_based:
    # The maximum distance from a marker to a wall to be considered part of the room
    marker_wall_distance_thresh: 6.0

  # Parameters for skeleton voxblox free-space room segmentation
  skeleton_based:
    # The minimum number of points needed to form a cluster
    min_cluster_vertices: 5
    # The maximum distance from a point of a cluster to a wall to be considered part of the room
    cluster_point_wall_distance_thresh: 1.5
    # The maximum distance from the cluster centroid to a wall centroid to be considered part of the room
    cluster_centroid_wall_centroid_distance_thresh: 5.0

  # Parameters for GNN-based room segmentation
  gnn_based:
    # Version of the GNN-based room segmentation to use
    gnn_version: 1 # 1: Older version used in S-Graphs, 2: Newer version used in VS-Graphs