Introduction

We are given skewed images with some degree. We are predicting the angle of skewed image using CNN based model then rotating it in opposite direction to get unskewed image.

How to run

python main.py 'image_fol/image_name.jpg' 
python main.py 'image_name.jpg' 

Training

python train.py

Test

python test.py

Important points

1) Change the k value to increase training dataset.Currently, we have used 
   only two random rotated data per image.
2) We are doing multi-class classification and there are 360 classes for each degree.
3) First, predicting rotated angle using this model then rotating the given image.

Skewed Image

Output Image