This directory contains scripts to generate tangent images, demo some non-deep learning applications, and highlight some other functionality and analysis. These example scripts should help clarify how to use this library.
- canny_edge_detection.py: Demonstrates the low-distortion benefits of tangent images for low-level tasks (specifically Canny edge detection)
- compare_sift_keypoints.py: Detects SIFT keypoints on both an equirectangular image (red) and tangent images (blue) and visualizes a comparison between them.
- compute_angular_resolution.py: Prints the FOV and angular resolution at each base level for the provided input resolution.
- create_tangent_image_obj.py: Writes an OBJ file of tangent images textured with the data from an equirectangular image
- draw_sift_keypoints.py: Detects SIFT keypoints on both an equirectangular image and tangent images with base levels {0, 1, 2}, and draws the results with scale and orientation. Saves the outputs to PDF files.
- generate_tangent_images.py: Turns an equirectangular images into tangent image patches. Also returns the binary mask of the valid (i.e. non-padding) regions of each tangent image.
- normalize_camera.py: Demonstrates the camera normalization used for pre-processing during the transfer learning experiments
- visualize_icosphere_sampling.py: Visualizes tangent images alongside the icosahedron as well as the sampling points on the sphere