I am talking about images in particular. If I collect a dataset for non-profit purposes (research) and an image is publicly available for anyone on the web (i.e. I can see it in google image search results, and the access to corresponding web page is not limited), does it mean that I can use it, if I mention source?
For example: 3d models themselves might be not available and cost money, but their previews are freely accessible by anyone on the web. Can I use these images as data for research and distribute an archive with these images freely? (including text file with authors of these images and URLs to pages!) Images themselves are available for free.
Or, I might collect a list of these URLs and publish a large txt file with these URLs so that everyone could download them if they need. That should not be a violation, is it? If I one downloads images to train an algorithm?
I feel like that is somewhat different from using images for educational or illustrative purposes.