Hard to download
Fantastic dataset but i found it was really hard to download the images from the URL. Is there something wrong with the way I downloaded?
How are you downloading and is there anything specific you're encountering?
There are various sizes available, here's the reference, bandwidth and storage requirements can be reduced by choosing a size that matches your target resolution rather than resizing a larger or original version.
If you're encountering 429 errors add delays and reduce concurrency.
Also check out img2dataset.
I wrote some codes by py to download images.
I firstly found the URLs of best quality from the database you provided and used requests them. But under this strategy, I quickly faced the issue of IP blocking.
Also, sometimes the website returns images that are 93 or 117 bytes in size, which are corrupted.
I’m now trying to use img2dataset, but the tool is a bit complex, and I’m still learning how to use it.
Anyway, thank you so much for the resources and help you’ve provided.
It seemed that I started img2dataset, but still could not download correctly, which liked:
0it [00:00, ?it/s]File sharded in 100 shards
Downloading starting now, check your bandwidth speed (with bwm-ng)your cpu (with htop), and your disk usage (with iotop)!
impossible to download on a single machine