[ Date Index ] [ Thread Index ] [ <= Previous by date / thread ] [ Next by date / thread => ]
On 24/02/18 15:18, mr meowski wrote:
On 24/02/18 14:56, Tom via list wrote:You may even find you can use AI for tagging photos too which, given the enourmous amount of photos some people take , may be worth the effort.If anyone wants to go this route there's always Google Photos - you can upload an unlimited* amount of photos to Google Drive for an online backup/repository accessible from anywhere. And of course their AI is the best in this business and will happily set to work recognising and classifying all your pics from EXIF data and contextual clues. Downside? Well, you have just uploaded all your photos to Google of course :| A lot of my clients use Google Photos and quite frankly they seem to love it. I don't use it personally (I'm not much of a photographer and rarely use my camera phone for anything) and I'm wary of cloud services in general but Google Photos does seem pretty cool, especially for a freebie. Cheers * unlimited if you don't mind them doing smart re-sampling to keep the sizes down a little, otherwise it's normal Google Drive storage rates to upload the original RAWs or whatever at full size.
From what I've seen it should be possible to do this at home with your own photos and no feeding the dragon: I've seen a video of someone doing RT object recognition on a Pi! and a browse around this site
http://caffe.berkeleyvision.org/ is quite interesting and you can download pretrained models: http://nbviewer.jupyter.org/github/BVLC/caffe/blob/master/examples/00-classification.ipynb
You dont need a TItan (but if anyones got when tell me where you live and leave the door on the latch ....
Tom te tom te tom -- The Mailing List for the Devon & Cornwall LUG https://mailman.dclug.org.uk/listinfo/list FAQ: http://www.dcglug.org.uk/listfaq