Step CV-Pipeline: data_prep

This CV-Pipeline component is designed for processing the dataset: checking and cleaning the dataset, transforming the markup, dividing dataset samples into train, valid and test, reviewing and processing data.

Input data for step CV-Pipeline: data_prep

• coco_datasets_images
  images of the downloaded dataset
• coco_datasets_annotations
  annotation files of the downloaded dataset

The output of this step CV-Pipeline is

• coco_train_dataset
  training dataset
• coco_eval_dataset
  validation dataset
• coco_test_dataset
  test dataset

How to run a step CV-Pipeline: data_prep

clone the repository: data_prep

git clone --recurse-submodules https://github.com/4-DS/obj_detect_binary-data_prep.git
cd obj_detect_binary-data_prep

run step CV-Pipeline:data_prep

python step.dev.py

or

step.prod.py