ml.cv

ml.cv#

Description#

Plugin to train and make computer vision predictions using machine learning models.

Also make sure that your OpenCV installation comes with the dnn module. To test it:

>>> import cv2.dnn

Configuration#

ml.cv:
    # No configuration required

Dependencies#

pip

pip install numpy opencv-python

Alpine

apk add py3-opencv py3-numpy

Debian

apt install python3-opencv python3-numpy

Fedora

yum install python-numpy python-opencv

Arch Linux

pacman -S python-numpy python-opencv

Actions#

Module reference#

class platypush.plugins.ml.cv.MlCvPlugin(**kwargs)[source]#

Bases: Plugin

Plugin to train and make computer vision predictions using machine learning models.

Also make sure that your OpenCV installation comes with the dnn module. To test it:

>>> import cv2.dnn
__init__(**kwargs)[source]#
predict(img, model_file, classes=None, resize=None, color_convert=None)[source]#

Make predictions for an input image using a model file. Supported model formats include all the types supported by cv2.dnn (currently supported: Caffe, TensorFlow, Torch, Darknet, DLDT).

Parameters:

  • model_file – Path to the model file

  • img – Path to the image

  • classes – List of string labels associated with the output values (e.g. [‘negative’, ‘positive’]). If not set then the index of the output neuron with highest value will be returned.

  • resize – Tuple or list with the resize factor to be applied to the image before being fed to the model (default: None)

  • color_convert – Color conversion to be applied to the image before being fed to the model. It points to a cv2 color conversion constant (e.g. cv2.COLOR_BGR2GRAY) and it can be either the constant value itself or a string (e.g. ‘COLOR_BGR2GRAY’).

Platypush documentation