Open jonochang opened 4 years ago
Hi @jonochang - try contacting help@pawsey.org.au. If this is unsuccessful please let me know and I'll see if I can sort an alternate download location.
Thanks for the prompt response, I've sent them an email to see if they support the download of large datasets.
@jonochang We don't seem to get good download speeds on the dataset either. I found that not all images may be needed during early testing. I wrote a few scripts to extract frames with multiple annotations from the metadata, download those selectively and generate a JSON file to be used in training. Perhaps you'll find a use for it too.
Hi, I tried to download your new annotations but the data provider PAWSEY shows offline for all files. Maybe this is just a temporary issue but I wanted to let you know, so you can check.
Edit: I noted this isn't true for all files but for these newer ones: https://data.pawsey.org.au/public/?path=/FDFML/labelled/measurementfiles https://data.pawsey.org.au/public/?path=/FDFML/labelled/speciesboxes
Thanks, @hoalarious , your download script is a good decision. I've improved it a bit - added multithreading download and safe download crash processing. Running it with param --threads=500 makes download really fast. Modified script attached download_dataset_from_json.py.md .
Hi!
I'm trying to download the frames from https://data.pawsey.org.au/public/?path=/FDFML/frames as a zip and it appears to die part way through. Is there a mirror of the dataset elsewhere, or is there a mechanism to resume downloads?