oakutils.blobs package

Subpackages

• oakutils.blobs.definitions package
  • Subpackages
    • oakutils.blobs.definitions.utils package
      • Submodules
      • oakutils.blobs.definitions.utils.conversion module
      • oakutils.blobs.definitions.utils.types module
      • Module contents
  • Submodules
  • oakutils.blobs.definitions.abstract_model module
  • oakutils.blobs.definitions.closing module
  • oakutils.blobs.definitions.dilation module
  • oakutils.blobs.definitions.erosion module
  • oakutils.blobs.definitions.gaussian module
  • oakutils.blobs.definitions.gftt module
  • oakutils.blobs.definitions.harris module
  • oakutils.blobs.definitions.hessian module
  • oakutils.blobs.definitions.laplacian module
  • oakutils.blobs.definitions.laserscan module
  • oakutils.blobs.definitions.opening module
  • oakutils.blobs.definitions.point_cloud module
  • oakutils.blobs.definitions.sobel module
  • Module contents
• oakutils.blobs.models package
  • Subpackages
    • oakutils.blobs.models.shave1 package
      • Module contents
    • oakutils.blobs.models.shave2 package
      • Module contents
    • oakutils.blobs.models.shave3 package
      • Module contents
    • oakutils.blobs.models.shave4 package
      • Module contents
    • oakutils.blobs.models.shave5 package
      • Module contents
    • oakutils.blobs.models.shave6 package
      • Module contents
  • Module contents
    • Note:
    • Attributes:

Module contents

Module for compiling, running, and creating custom models.

Note:

This module requires the [compiler] dependencies to be installed. If not only the models submodule will be available.

Submodules

definitions

Contains the definitions for the models.

models

Contains the pre-compiled models.

Classes

BenchmarkData

Dataclass for storing benchmark data.

Metric

Dataclass for storing metrics.

Functions

bencmark_blob

Benchmark a blob.

compile_model

Compiles a model.

compile_onnx

Compiles an onnx model into a blob.

clear_cache

Clears the cache of compiled blobs.

get_blob

Load a blob from a path.

get_cache_dir

Get the cache directory path.

get_input_layer_data

Get the input layer data for a blob.

get_output_layer_data

Get the output layer data for a blob.

get_layer_data

Get the input and output layer data for a blob.

get_model_name

Get the name of a compiled model file.

get_model_path

Get the path to the model blob.

class oakutils.blobs.BenchmarkData(latency: 'Metric')

Bases: object

latency: Metric

class oakutils.blobs.Metric(mean: 'float', std: 'float', min: 'float', max: 'float')

Bases: object

max: float

mean: float

min: float

std: float

oakutils.blobs.benchmark_blob(blob: Path | str, iterations: int = 100, warmup_iterations: int = 10, *, is_yolo: bool | None = None, warmup: bool | None = None) → BenchmarkData

Benchmark a blob.

Parameters:

• blob (Path | str) – The path to the blob file or directory containing the blob files.

• iterations (int, optional) – The number of iterations to run, by default 100

• warmup_iterations (int, optional) – The number of warmup iterations to run, by default 10

• is_yolo (bool, optional) – Whether the blob is a Yolo blob, by default None. If is_yolo is True, then the blob_path should be a directory and the directory should contain a .json file generated during compilation of the blob.

• warmup (bool, optional) – Whether to run warmup iterations, by default None. If warmup is True, then the warmup_iterations will be run before the iterations.

Returns:

The benchmark data.

Return type:

BenchmarkData

Raises:

• FileNotFoundError – If the blob file or directory does not exist.

• ValueError – If is_yolo is True and the blob_path is not a directory.

• ValueError – If the directory does not contain exactly one blob file.

• ValueError – If the directory does not contain exactly one json file.

oakutils.blobs.get_blob(blob_path: Path | str) → Blob

Load a blob from a file.

Parameters:

blob_path (Path | str) – The path to the blob.

Returns:

The blob object.

Return type:

dai.OpenVINO.Blob

Raises:

FileNotFoundError – If the blob does not exist.

oakutils.blobs.get_input_layer_data(blob: Path | str | Blob) → list[LayerData]

Get the input layer data for a blob.

Parameters:

blob (Path | str | dai.OpenVINO.Blob) – The path to the blob.

Returns:

A list of LayerData objects.

Return type:

list[LayerData]

oakutils.blobs.get_layer_data(blob: Path | str | Blob) → tuple[list[LayerData], list[LayerData]]

Get the input and output layer data for a blob.

Parameters:

blob (Path | str | dai.OpenVINO.Blob) – The path to the blob.

Returns:

A tuple of lists of LayerData objects.

Return type:

tuple[list[LayerData], list[LayerData]]

oakutils.blobs.get_model_path(model_type: str, model_attributes: list[str], shaves: int) → Path

Get the path to the model blob.

Parameters:

• model_type (str) – The model type to get the path for. Examples include: [‘gaussian’, ‘sobel’]

• model_attributes (list[str]) – The model attributes to get the path for. An example could be [‘15’] for a gaussian model using a 15x15 kernel size.

• shaves (int) – The number of shaves the model was compiled for.

Returns:

The path to the model blob.

Return type:

Path

Raises:

• FileNotFoundError – If the returned model blob does not exists.

• ValueError – If no model blob paths could be formed from the attributes and shaves.

oakutils.blobs.get_output_layer_data(blob: Path | str | Blob) → list[LayerData]

Get the output layer data for a blob.

Parameters:

blob (Path | str | dai.OpenVINO.Blob) – The path to the blob.

Returns:

A list of LayerData objects.

Return type:

list[LayerData]