Vincentqyw
update: features and matchers
a80d6bb

A newer version of the Gradio SDK is available: 4.37.2

Upgrade

HPatches Sequences / Image Pairs Matching Benchmark

Please check the official repository for more information regarding references.

The dataset can be downloaded by running bash download.sh - this script downloads and extracts the HPatches Sequences dataset and removes the sequences containing high resolution images (> 1600x1200) as mentioned in the D2-Net paper. You can also download the cache with results for all methods from the D2-Net paper by running bash download_cache.sh.

New methods can be added in cell 4 of the notebook. The local features are supposed to be stored in the npz format with three fields:

  • keypoints - N x 2 matrix with x, y coordinates of each keypoint in COLMAP format (the X axis points to the right, the Y axis to the bottom)

  • scores - N array with detection scores for each keypoint (higher is better) - only required for the "top K" version of the benchmark

  • descriptors - N x D matrix with the descriptors (L2 normalized if you plan on using the provided mutual nearest neighbors matcher)

Moreover, the npz files are supposed to be saved alongside their corresponding images with the same extension as the method (e.g. if method = d2-net, the features for the image hpatches-sequences-release/i_ajuntament/1.ppm should be in the file hpatches-sequences-release/i_ajuntament/1.ppm.d2-net).

We provide a simple script to extract Hessian Affine keypoints with SIFT descriptors (extract_hesaff.m); this script requires MATLAB and VLFeat.

D2-Net features can be extracted by running:

python extract_features.py --image_list_file image_list_hpatches_sequences.txt