ODToolkit 0.0.6

An object detection toolkit for data mining and visualization

Object_Detection_Toolkit

A tool kit for checking boxes in a dataset and visualize predictions' distribution

Install

To install this tool, simply

pip install Object_Detection_Toolkit

Functions

There are few functions in the Visualizer object.

1. BoxVisualizer.show_boxes(**kargs): show all the boxes in a dataset, including there actual positions, histogram and heatmaps

2. BoxVisualizer.get_fpfn(**kargs): calculate FP, FN and correct predictions giving GT and Pred

3. BoxVisualizer.show_fpfn(**kargs): show FP, FN and correct boxes obtained from step 2

Usage

To use, simply follow belows snippet:

from Object_Detection_Toolkit.Visualizer import BoxVisualizer

bv = BoxVisualizer(img_w=640, img_h=480)    # image size of your dataset

train_lbl = [...]                           # list of your train label txt
test_lbl = [...]                            # list of your test label txt
bv.show_boxes(labels=[train_lbl, test_lbl], 
              mode=['train', 'test'])       # show the boxes in a dataset

FP, FN, correct = bv.get_fpfn(**kargs)      # calculate FP, FN and Correct boxes
bv.show_fpfn(FP, FN, correct)               # show FP, FN and Correct boxes calculated just now

Unit-test

For unit-test, we use Behave, which provides a very flexible environment for all kinds of tests

To view a sample, go to unit_tests folder and check *.feature and steps/*.py files

To run tests, simply:

behave unit_tests/               # run all tests
behave unit_tests/xxx.feature    # run particular test you want

To contribute

If you want to contribute to this project, please submit a pull request with unit tests.