5 ms·
Let's say I did a survey of images on the internet, using a crawler to look for and download each image. I make counts of how many of each file format I find, t
by Canada 3y ago
Let's say I did a survey of images on the internet, using a crawler to look for and download each image. I make counts of how many of each file format I find, their sizes and color depths, and a hash to count dupes. I don't save or redistribute any image. I let the public buy copies of the statistics I have compiled.
Can a copyright owner claim infringement, and demand that I update my statistics to make it as if my crawler didn't find his images?
- Kim_Bruning 3y agoAlso, your resulting data-set would be at least an order of magnitude larger than a Stable Diffusion model.
- archontes 3y agoAlso, I believe this is irrelevant.
- Kim_Bruning 3y agoPretty much! To the actual case at hand it probably is irrelevant. People do use the LAION dataset for something of course. Originally maybe just for image recognition models, and now of course for image generation models.
- klyrs 3y agoThis is disingenuous. Large models are observed to reproduce works from their training data, sometimes verbatim, often close enough to constitute a derived work. You scraping metadata would never be able to do that.
- marssaxman 3y agoThe dataset in question is not a model, and contains no images; it is a long list of URLs, with associated tags. Someone who wished to train a model could use this information to download a lot of pictures from various places on the internet, then create training and validation sets; but LAION themselves do not provide images or model weights. The photographer went after the wrong target here; if he wants his images taken off the internet he should contact whoever is hosting them at some publicly accessible URL.
- archontes 3y agoA model being able to produce an infringing work doesn't mean the model is an infringing work. I can produce work myself that is infringing or not infringing as a result of learning.
- Canada 3y agoI don't think your questioning of my sincerity is helpful. I agree that generative models have been observed to reproduce training input verbatim, but that's not what I understand to be at issue in the article. The issue here is whether you can presume the consumer of a list will do something illegal with an item on the list without any evidence, and therefore threaten the compiler of the list.