Built-in Data Types

PoSDK provides the following built-in DataIO data types for managing storage and serialization of common SfM data:

Built-in Data Type	Core Data Type	Save/Load Supported Formats	Supports Auto Evaluation
data_images	ImagePaths	folder	✗
data_camera_models	CameraModels	.pb	✗
data_feature	ImageFeatureInfo*	.pb	✗
data_features	FeaturesInfo	.pb	✗
data_matches	Matches	.pb	✗
data_bearing_pairs	BearingPairs	.pb	✗
data_tracks	Tracks	.pb,.tracks	✗
data_relative_rotations	RelativeRotations	-	✗
data_relative_poses	RelativePoses	.g2o, .pb	✓
data_global_rotations	GlobalRotations	.rot	✗
data_global_translations	GlobalTranslations	-	✗
data_global_poses	GlobalPoses	.po, .g2o, .rot	✓
data_points_3d	WorldPointInfo	.txt, .ply (+.ids)	✗

Note

• Save/Load Supported Formats:

  • .pb - Protobuf binary format (high performance, cross-platform)

  • .txt - Text format (human-readable)

  • .ply - PLY point cloud format (only for 3D point data)

  • +.ids - PLY format usage status auxiliary file

• Auto Evaluation: Data types marked with ✓ have overloaded Evaluate() function, supporting automatic comparison evaluation with ground truth data

• Core Data Type Detailed Description: Please refer to Appendix A: PoSDK Core Data Types

Common SfM Data Types

PoSDK’s type system adopts modular design, distributing common SfM data types across multiple header files for easier maintenance. These types include:

• Basic Types: Identifiers, matrix vectors, smart pointers, etc.

• Camera System: Camera models, intrinsics, distortion parameters

• Feature System: Feature points, descriptors, image feature management

• Match System: Feature matching, viewing vectors

• Track System: Observation information, track management

• Pose System: Relative poses, global poses, pose formats

• 3D Point Cloud: World coordinate points, point cloud management

Detailed Description: For complete definitions, member variables, available methods, and usage examples of all common SfM data types, please refer to Appendix A: PoSDK Core Data Types.