MicrostockGroup Sponsors


Author Topic: Shutterstock rejecting clips and photos like crazy for no good reason recently.  (Read 2034 times)

0 Members and 1 Guest are viewing this topic.

« Reply #25 on: November 26, 2019, 21:01 »
0
...
It's hurting my workflow because I use to submit several pictures together when I can use the same keywords (f.i. several views of a building in a small series, and then I finish with that building). ...

aren't you using metadata? then it wouldnt matter whether you submit one at a time or all in a batch


« Reply #26 on: November 27, 2019, 02:56 »
+1
...
It's hurting my workflow because I use to submit several pictures together when I can use the same keywords (f.i. several views of a building in a small series, and then I finish with that building). ...

aren't you using metadata? then it wouldnt matter whether you submit one at a time or all in a batch

I mean that I use to select files by groups. Now I'm always thinking in the SS problem when I revise my pictures to collect my batches.

I use text files and copy-paste.

« Reply #27 on: December 02, 2019, 09:03 »
+1
congrats Shutterstock, it's nice to finally be able to realize something after so long time. but... what about duplicated pictures were accepted for 15 years?
« Last Edit: December 02, 2019, 09:09 by madman »

« Reply #28 on: December 02, 2019, 09:30 »
0
It's all about AI...i don't think any human being can be so inconsistent and stupid on curation...
I don't believe AI can be inconsistent. It is not one reviewer. Seems to me the quality control on reviewers is broken. If reviewers receive next to no training or supervision and are required to review at a ridiculous pace I wouldn't be surprised if some go bat dropping crazy.

Because you do not know how AI works. AI should be consistent in a real world but especially when it is still learning, it is not. If you have trained it on million images, it is still posiible that even one additional image incorporated to the learning can break what was already working (change the decision for images which have been on the borderline and they do not even have to have anything in common with the new one because of all the dependencies - one changed weight in the system will influence other...).

Uncle Pete

  • Great Place by a Great Lake - My Home Port
« Reply #29 on: December 02, 2019, 16:05 »
0
It's all about AI...i don't think any human being can be so inconsistent and stupid on curation...
I don't believe AI can be inconsistent. It is not one reviewer. Seems to me the quality control on reviewers is broken. If reviewers receive next to no training or supervision and are required to review at a ridiculous pace I wouldn't be surprised if some go bat dropping crazy.

Because you do not know how AI works. AI should be consistent in a real world but especially when it is still learning, it is not. If you have trained it on million images, it is still posiible that even one additional image incorporated to the learning can break what was already working (change the decision for images which have been on the borderline and they do not even have to have anything in common with the new one because of all the dependencies - one changed weight in the system will influence other...).

Quite the opposite, AI relies on many results and the weight of those, plus the training, to make a decision, not one wrong results that changes the whole system. If one wrong entry changed everything, that would be a terribly flawed system.

In simpler terms, a machine learns by looking for patterns among massive data loads, and when it sees one, it adjusts the program to reflect the truth of what it found. The more data you expose the machine to, the smarter it gets. And when it sees enough patterns, it begins to make predictions.

The key is more data, not individual data.

I know how much everyone likes blaming AI or computers or some conspiracy, something outside of, as simple as, the reviewers make mistakes, or possibly the reviewers are doing what they are told to do. Yes, there's additional software to assist and analyze, but that's not AI. I know how many people say, "I'm right, the reviewer is an idiot" and get defensive. Of course, that stupid AI can't see how perfect our images.  :)

And before someone says they have had SS rejected images, accepted at XYZ, I've had Alamy rejected and Adobe rejected images, accepted at SS. Even stranger was one rejected at Adobe that is my best selling on DT (which is hardly saying much), but still, even between different agencies we get different reviews.
« Last Edit: December 02, 2019, 16:10 by Uncle Pete »

« Reply #30 on: December 03, 2019, 11:12 »
0
It's all about AI...i don't think any human being can be so inconsistent and stupid on curation...
I don't believe AI can be inconsistent. It is not one reviewer. Seems to me the quality control on reviewers is broken. If reviewers receive next to no training or supervision and are required to review at a ridiculous pace I wouldn't be surprised if some go bat dropping crazy.

Because you do not know how AI works. AI should be consistent in a real world but especially when it is still learning, it is not. If you have trained it on million images, it is still posiible that even one additional image incorporated to the learning can break what was already working (change the decision for images which have been on the borderline and they do not even have to have anything in common with the new one because of all the dependencies - one changed weight in the system will influence other...).
Possible yes likely not so much and it should be getting better on average not worse. The simpler explanation remains human failings.


 

Related Topics

  Subject / Started by Replies Last post
21 Replies
7354 Views
Last post April 06, 2009, 06:03
by Magnum
199 Replies
58972 Views
Last post October 28, 2011, 19:15
by Mantis
27 Replies
5994 Views
Last post July 31, 2013, 16:42
by Jo Ann Snover
11 Replies
4694 Views
Last post September 27, 2016, 03:27
by skysense
13 Replies
2206 Views
Last post January 28, 2019, 17:09
by Microstockphoto

Sponsors

Microstock Poll Results