The Required Datasets for Training.

[zhenpengju]

As indicated on GitHub, training our own model(superpoint-lightglue) requires two steps and a total of two datasets: 1(1M-image distractor set of the Oxford-Paris retrieval dataset.) and 2(MegaDepth dataset). However, I am not sure which part of these two datasets I should download. Is there anyone who can share the download links and tell me which parts to download?

[yuegao998]

The dataset code have the dataset message you need.you can check in the code and fine the dataset url and file.