package scenenet

Get desktop application:
View/edit binary Protocol Buffers messages

Used in: Trajectory

• optional int32 instance_id = 1

• optional string semantic_wordnet_id = 2

• optional string semantic_english = 3

• optional Instance.InstanceType instance_type = 4

• optional LightInfo light_info = 5

  This information is only filled in for the respective type

• optional RandomObjectInfo object_info = 6

Used in: Instance

• BACKGROUND = 1

  This is the instance type when no object is present, e.g. because of looking out a window into nothingness

• LAYOUT_OBJECT = 2

  This is an object that is hard coded into the layout and does not move. This type does not have a transformation or shapenet hash

• LIGHT_OBJECT = 3

  This is a randomly positioned light source

• RANDOM_OBJECT = 4

  This means the object is a randomly positioned shapenet object. The object has a transformation and scale parameter in the object_info variable.

Used in: Instance

• optional LightInfo.LightType light_type = 1

• optional Power light_output = 2

  Light intensity

• optional Position position = 3

  This is the center for sphere type lights. And corner for others

• optional float radius = 4

  This is only for SPHERE lights

• optional Position v1 = 5

  This is only for PARALLELOGRAM lights

• optional Position v2 = 6

Used in: LightInfo

• SPHERE = 1

• PARALLELOGRAM = 2

Used in: View

• optional Position camera = 1

  The position of these two points define the camera view. The y vector is defined as [0,1,0]. For an example of how to calculate the camera view coordinate system, see the python codebase.

• optional Position lookat = 2

• optional float timestamp = 3

Used in: LightInfo, Pose

• optional float x = 1

• optional float y = 2

• optional float z = 3

Used in: LightInfo

• optional float r = 1

• optional float g = 2

• optional float b = 3

Used in: Instance

• optional string shapenet_hash = 1

• optional float height_meters = 2

• optional RandomObjectInfo.Transformation object_pose = 3

  The transformation gives the transformation applies to an object, about the center of the base plane of its axis-aligned bounding box.

Used in: RandomObjectInfo

• optional float translation_x = 1

  The 3x4 matrix is as follows: rotation_mat11 rotation_mat12 rotation_mat13 translation_x rotation_mat21 rotation_mat22 rotation_mat23 translation_y rotation_mat31 rotation_mat32 rotation_mat33 translation_y

• optional float translation_y = 2

• optional float translation_z = 3

• optional float rotation_mat11 = 4

• optional float rotation_mat12 = 5

• optional float rotation_mat13 = 6

• optional float rotation_mat21 = 7

• optional float rotation_mat22 = 8

• optional float rotation_mat23 = 9

• optional float rotation_mat31 = 10

• optional float rotation_mat32 = 11

• optional float rotation_mat33 = 12

Used in: Trajectory

• optional SceneLayout.LayoutType layout_type = 1

• optional string model = 2

  This is the name of the SceneNet model used for the layout

Used in: SceneLayout

• BATHROOM = 1

• BEDROOM = 2

• KITCHEN = 3

• LIVING_ROOM = 4

• OFFICE = 5

• repeated Trajectory trajectories = 1

  This is the root list which stores all of the available trajectories

Used in: Trajectories

• optional SceneLayout layout = 1

• repeated Instance instances = 2

  The first instances[0] is always the 'background' and undefined class when for example looking out windows

• repeated View views = 3

  These are ordered sequentially for a trajectory

• optional string render_path = 4

  This stores the path from the root data directory to the trajectory data folder. If the trajectories are stored as: /path/i/extracted/{val/train}/0/123/photo/0.jpg then this path will be '0/123' designating the trajectories folder

Used in: Trajectory

• optional int32 frame_num = 1

  These increment by the number of skip frames, i.e. 0,25,50...7475.

• optional Pose shutter_open = 2

  The photo is rendered by integrating uniformly sampled exposures between the following two poses

• optional Pose shutter_close = 3