Utils: pyvar.ml.utils

Label Reader Class

pyvar.ml.utils.label

platform

Unix/Yocto

synopsis

Class to handle labels file from Machine Learning models.

class pyvar.ml.utils.label.Label(labels_file_path=None)

Variables

• labels_file – path to the labels file.

• list – list with the labels from the label file.

read_labels(category=None)

Reads Machine Learning labels file and save the result as a list.

Parameters

category (str) – model category (classification or detection).

Returns

True if the labels file was read successfully.

Overlay Class

pyvar.ml.utils.overlay

platform

Unix/Yocto

synopsis

Class to handle overlay on single images and frames.

class pyvar.ml.utils.overlay.Overlay

Variables

• inference_time_info – shows the inference time on image/frame;

• scores_info – shows the scores on image/frame;

• extra_info – shows extra info on image/frame;

• framerate_info – shows framerate on image/frame.

static generate_colors(labels)

info(category=None, image=None, top_result=None, labels=None, inference_time=None, model_name=None, source_file=None, fps=None)

Draw information on single images and frames such as inference time, scores, model name, and source file.

Parameters

• category (str) – model category (classification or detection);

• image (numpy array) – original image to overlay the information.

• top_result (list) – top results from the inference.

• labels (list) – list of the read labels.

• inference_time (str) – inference time from TFLiteInterpreter class.

• model_name (str) – the model name.

• source_file (str) – the source file name.

• fps (float) – fpsit from Framerate class.

Returns

The numpy.array image format with the overlaid information.

Frames Per Second Calculator Class

pyvar.ml.utils.framerate

platform

Unix/Yocto

synopsis

Class to calculate framerate from videos and real time cameras devices.

class pyvar.ml.utils.framerate.Framerate

fpsit()

Calculates the frames per second and save it in the fps attribute.

File Transfer Protocol Class

pyvar.ml.utils.retriever

platform

Unix/Yocto

synopsis

Class to retrieve packages from Variscite FTP server.

class pyvar.ml.utils.retriever.FTP(host=None, user=None, passwd=None)

This class can be used as reference only. It is not for production-ready.

retrieve_package(package_dir=None, package_filename=None, category=None)

Retrieve package from the FTP server.

Parameters

• package_dir (str) – package directory;

• package_filename (str) – model package file name;

• category (str) – model category (classification or detection).

Returns

True if the package file was downloaded successfully. False if not.

Resizer Class

pyvar.ml.utils.resizer

platform

Unix/Yocto

synopsis

Class to resize image/frame.

class pyvar.ml.utils.resizer.Resizer

Variables

• frame – original frame;

• frame_resized – resized frame;

• image – original image;

• image_path – image path;

• image_resized – image resized;

• model_height – model height;

• model_width – model width.

resize_frame(frame=None, width=None, height=None, expand_dims=True)

Resize the frame.

Parameters

• frame (numpy array) – original frame to be resized;

• width (int) – new width to resize;

• height (int) – new height to resize;

• expand_dims (bool) – if expand dimensions or not

Returns

The numpy.array resized frame is stored at the frame_resized attribute.

resize_image(image_path=None, width=None, height=None, expand_dims=True)

Resize the image.

Parameters

• image_path (numpy array) – original image to be resized;

• width (int) – new width to resize;

• height (int) – new height to resize;

• expand_dims (bool) – if expand dimensions or not

Returns

The numpy.array resized image is stored at image_resized attribute.

set_sizes(engine_input_details=None)

Set the height and width model size.

Timer Calculator Class

pyvar.ml.utils.timer

platform

Unix/Yocto

synopsis

Python Class to calculate time.

class pyvar.ml.utils.timer.Timer

timeit()

Calculates the time and save it in the time attribute.