Does anyone know of a good computer vision object recognition training dataset of landmarks (specific buildings, specific natural wonders, etc) that has bounding boxes specified for each landmark?

There's the Google-Landmarks Dataset, which has exactly what I need...but three years ago, they deleted all their data from it! While, unfortunately, v2 doesn't seem to have bounding box information

Ideally, there will be a .yaml I can use in training, but... If not... I can always just make it myself