Image Augmentation for Object Detection

Images augmentation for Object detection tasks

Getting Started

This pipeline can use to Augment data in CSV Annotation Format

(CSV Annotation) Format:

Make sure that, your csv file in this order.

Installation

Clone this repo with:

git clone git@github.com:LahiRumesh/Object-Detection_Data-Augmentation.git
cd Object-Detection_Data-Augmentation/

Install required packages :

pip install -r requirements.txt

Image Augmentation

Make changes to user_inputs.ini ,

[USER_DATA]
IMAGE_FOLDER=images   # Folder path which contain images
INPUT_CSV=train.csv   # CSV file path with 
AUGMENT_FOLDER=image_output  # Images save into this directory after augmentaion
AUGMENT_CSV=output.csv  # CSV file which contain augmentation values

If you want to Rotate Image set ROTATE_IMAGES=True

[Random_Rotate]
ROTATE_IMAGES=True 
ROTATE_VALUE=45

Further more If you want to Rotate and Resize the images in same time, set ROTATE_IMAGES=True & RESIZE_IMAGES=True

[Random_Rotate]
ROTATE_IMAGES=True 
ROTATE_VALUE=45

[Resize]
RESIZE_IMAGES=True 
RESIZE_VALUE=416

Make sure to Select atleast one Augmentation Type, Make your changes and save user_inputs.ini file

python augmentation.py

Acknowledgements

Data Augmentation For Object Detection for more information.