Toy experiment to compare Conventional and Deep Image Model for Homography estimation

This code is a assignment for Intelligence Integrated System class, April 2020

This code is writen by jupyter notebook with keras and OpenCV:

• Deep Image model: reimplemented from [DeTone D, Malisiewicz T, Rabinovich A. Deep image homography estimation[J]. arXiv preprint arXiv:1606.03798, 2016.] https://arxiv.org/abs/1606.03798.
  You can get a pretrained model from here:https://drive.google.com/open?id=0B88aZDadyYr8TTBSY3JvYTVKVnc

• And some conventional methods in OpenCV library

Dependencies

• OpenCV 3.4.2
• Keras
• TensorFlow

References

• https://github.com/yishiliuhuasheng/deep_image_homography_estimation
• https://www.pyimagesearch.com/2016/01/11/opencv-panorama-stitching/