Enums¶

Enumeration types for MGT-python parameter validation.

Using StrEnum so that enum members compare equal to their string values, maintaining full backward compatibility with code that passes plain strings. All enumerations support case-insensitive construction:

>>> BlurType("average") == BlurType.AVERAGE
True
>>> BlurType("AVERAGE") == BlurType.AVERAGE
True

FilterType ¶

Bases: _MgEnum

Pixel-value filter applied to the frame-difference stream.

Attributes:

Name	Type	Description
`REGULAR`		Values below threshold are set to 0; values above are kept as-is.
`BINARY`		Values below threshold → 0; values above threshold → 255.
`BLOB`		Individual pixels are removed with an erosion filter.

BlurType ¶

Bases: _MgEnum

Spatial blur applied before the frame-difference computation.

Attributes:

Name	Type	Description
`NONE`		No blurring is applied.
`AVERAGE`		A 10 × 10 pixel box-blur is applied.

CropMode ¶

Bases: _MgEnum

Video cropping strategy.

Attributes:

Name	Type	Description
`NONE`		No cropping.
`MANUAL`		Opens an interactive window; the user draws a rectangle.
`AUTO`		Automatically detects the area of significant motion.

PoseModel ¶

Bases: _MgEnum

Pose estimation skeleton model.

Attributes:

Name	Type	Description
`BODY_25`		OpenPose BODY_25 dataset (25 keypoints).
`COCO`		COCO dataset (18 keypoints).
`MPI`		MPII dataset (15 keypoints).
`MEDIAPIPE`		Google MediaPipe Pose (33 landmarks).

PoseDevice ¶

Bases: _MgEnum

Compute backend for pose estimation inference.

Attributes:

Name	Type	Description
`CPU`		Run on CPU.
`GPU`		Run on GPU (CUDA / OpenCL).

DataFormat ¶

Bases: _MgEnum

Output data file format.

Attributes:

Name	Type	Description
`CSV`		Comma-separated values.
`TSV`		Tab-separated values.
`TXT`		Plain text (space-separated).

Note

JSON and HDF5 are reserved for future use and are not yet implemented by any processing function. Passing them will silently fall back to CSV. Use 'csv', 'tsv', or 'txt'.