King@lemy.lol to People Twitter@sh.itjust.works · 10 months agoPSA for artistslemy.lolimagemessage-square39fedilinkarrow-up1222arrow-down119file-text
arrow-up1203arrow-down1imagePSA for artistslemy.lolKing@lemy.lol to People Twitter@sh.itjust.works · 10 months agomessage-square39fedilinkfile-text
minus-squareDeceptichum@kbin.sociallinkfedilinkarrow-up32·edit-210 months agoI’m only vaguely familiar with ML datasets and have only trained on local data, but I’ve never heard of this? Can anyone provide some evidence this is the case? Edit: Looking further I can still only find datasets containing the image files, ex. https://www.lvisdataset.org/dataset https://www.v7labs.com/open-datasets
minus-squareYmmelbackwards@lemmy.worldlinkfedilinkarrow-up17·10 months agoLAION is one of the big dogs (https://laion.ai/). Their datasets consist of urls and metadata.
minus-squarewhoelectroplateuntil@sh.itjust.workslinkfedilinkarrow-up17·edit-210 months agodeleted by creator
minus-squareDeceptichum@kbin.sociallinkfedilinkarrow-up4arrow-down1·10 months agoAh perfect, thank you so much! https://github.com/rom1504/img2dataset Seems to be the main tool, I’ll have something new to explore this weekend.
I’m only vaguely familiar with ML datasets and have only trained on local data, but I’ve never heard of this? Can anyone provide some evidence this is the case?
Edit: Looking further I can still only find datasets containing the image files, ex.
https://www.lvisdataset.org/dataset
https://www.v7labs.com/open-datasets
LAION is one of the big dogs (https://laion.ai/). Their datasets consist of urls and metadata.
deleted by creator
Ah perfect, thank you so much!
https://github.com/rom1504/img2dataset
Seems to be the main tool, I’ll have something new to explore this weekend.