Seamless Panorama Stitching (MyAutopano)

Phase 1 : Geometrical computer vision

The code can be executed using the wrapper.py from the Phase1 folder.

1. Input Data: Sequence of images

2. Corner Detection and Adaptive Non-Maximal Suppression (ANMS):

3. Feature Matching

After RANSAC

4. Panorama Stiching

Phase2

The models can be trained using the train.py file and can be tested using the test.py file. Both the files default to the Supervised version of the algorithm and must be provided the modeltype argument to change to the unsupervised version. Similarly, both the files must be given the path for checkpoints.