-
Notifications
You must be signed in to change notification settings - Fork 1
Image Dataset Sources
NaskyD edited this page May 24, 2017
·
4 revisions
- Cityscapes Dataset: per-pixel labels with names(!) (-> BB extraction via script possible), has buildings category, needs registration and moderator approval
- Oxford Building Dataset: 5k images, semi-sorted, no BBs
- Oxford Paris Dataset: 500-1k images, sorted by landmark, no BBs
- Paris500k Dataset: 501356 images in total, cralled from flickr, 10 different building labels
- LabelMe: has per-pixel labels (-> BB extraction via script possible)
- MIT Scene Parsing Benchmark: semi-useful per-pixel labels (Buildings are often too big / cropped on images)
- Facade: This is one example of an facade dataset, but certainly not useful for us. The since nearly all of the image is covered in from the facade alone, the idea to use the per-pixel labels for implicit BB won't help much for building recognition among other scenery.