Skip to content

manushreegangwar/data_augmentor

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

6 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Data Augmentor

Problem Description

Given a series of images in multiple domains, such as RGB, depth and normal, create augmented datasets by applying transformations, such as rotation and blur, to images in each domain. For more details, see this link.

Setup

conda create -n AUG python=3.11

conda activate AUG

python setup.py develop

Running Augmentor

Before running the script, download the sample data from this link and save under data or a path of your choice.

python augmentor.py --config-file path/to/config

Data Augmentation framework

1. config

Config parameters are defined as a YAML file for augmentor

2. Dataset

Input images are loaded as data_augmentation.loader.Dataset

3. Transform

Image transforms currently supported are defined in data_augmentation.transforms

4. Augment

To apply a set of image Transform to input Dataset, use the data_augmentation.augment.Augment

Sample results

Go to sample_annotations directory for sample annotations generated using augmentor.py

References

  1. YACS
  2. cocoapi

About

Implementation of a data augmentation tool

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages