I submitted, with my Pokémon Go account ‘sup3rfm’, a 19th-century fountain-pond with a statue, which is situated in the gardens of Palácio Palmela (a Portuguese national monument). It was immediately rejected by the automated system (ML). Can this rejection be reversed? I’m worried that if I try submitting it again, it will be unfairly rejected in the same way.
Rejection: “This submission was declined by Niantic’s automated process for not adhering to Wayfarer criteria”
Title: O Lago Romântico do Jardim do Palácio Palmela Description: Este lago sereno, aninhado no Jardim do Palácio Palmela, evoca uma atmosfera de tranquilidade e romantismo. As suas margens irregulares, adornadas com rochas e vegetação rasteira, emolduram a superfície calma da água, que reflete a luz filtrada das árvores circundantes. Pequenos elementos escultóricos emergem da água e das margens, adicionando um toque de mistério e encanto ao cenário. Integrado na paisagem ajardinada, este lago convida à contemplação e ao descanso. Supplemental information: Critério: Excelente ponto de interesse no jardim do Palácio Palmela, da PGR.
Location: Lisbon, Portugal
Coordinates: 38.71965,-9.153783 Google Maps Main photo
Title: The Romantic Lake of the Palmela Palace Garden
Description: This serene lake, nestled within the Palmela Palace Garden, evokes an atmosphere of tranquility and romance. Its irregular banks, adorned with rocks and low-lying vegetation, frame the calm surface of the water, which reflects the filtered light of the surrounding trees. Small sculptural elements emerge from the water and the banks, adding a touch of mystery and charm to the scene. Integrated into the landscaped gardens, this lake invites contemplation and relaxation.
Supplemental information: Criterion: Excellent point of interest in the garden of the Palmela Palace, belonging to the Attorney General’s Office.
The ML probably got overwhelmed by the vegetation, especially with the reflection of the trees in the water. I actually like the supporting photo better, I think the pond shows more clearly although I’d crop the person out of it.
It does look potentially valid to me, if cropping the vegetation doesn’t work then you may need to take it to appeal.
You have two appeals, each on a 20 day counter before you can use them again, if you would like to try to get this rejection overturned.
I can’t see this area on street view to recommend a different angle which might more clearly be something that is potentially eligible to the ML (machine learning ai) model. In this main photo, I had trouble seeing that there is a statue in the pond. I agree that there is no point in resubmitting the same photo, but if you can get a picture where the pond and statue stand out more, maybe that will help get past the “automated process.” I can actually see the statue in the pond better in the supporting photo.
I understand that you want to keep the nomination for the pond because the statue in it would not have safe pedestrian access.
Yes crop out the other parts for the main photo so that pond dominates - the current picture is more about the setting of the pond than the pond itself.
I think if you get down low you might find an angle where the statue shows up better against the grass background. Its presence helps make the whole pond a feature
I really like your nomination, but it isnt very “ML friendly” like the others have mentioned.
I would try to get that fish statue centred and larger within the main picture, while still submitting the overall pool. This might help the ML understand what it’s seeing.
Your current picture has a lot of ground in front of the pool. I think the ML got confused with thinking the picutre is of a natural green space rather than a man made garden.
Otherwise I would just appeal. A human should be able to see the value in the submission pretty easily
Niantic’s ML/AI is incredible weak. I run past the imaged through different available and free to use AI, all of them described it as a pond, body of water, with the statue in the middle.
Thanks for the feedback.
I’ve heard about those appeals. Unfortunately, the appeals team has a poor track record, in my opinion:
This time, I find myself with a valid contribution unfairly declined by Niantic’s ML. Having to wait and spend another appeal because of their issues is exhausting.
My photo captures the whole pond, even the dry walled area where ducks usually stay (at the bottom of the picture). I could crop it, but it won’t capture the whole pond, as is.
The real focus of all this is: ML got it completely wrong, causing an additional issue for the voluntary submitted to solve. It’s not the right way to do it. It’s not fair.
We all saw it as a pond. Having to think about a submission and make it ML friendly is too much. We have to do that now? How about Niantic improve their own system? Unfortunately, the topics at the previous community platform were deleted, but I had a few threads about how poor the ML was – all decisions were manually reverted by Niantic.
Then increasing the risk of being ‘not safe for pedestrians’…
Don’t get me started on the appeals being a solution
Agree everyone could see a pond. But not everyone here said this clearly met criteria as a great place for exercise, exploration, or being social. It could easily be seen as not “distinct” and anything that you can do to make it look distinct would help it in review. I rarely see a pond in review I would accept.
You’re kidding, right? You’ve just ignored the description and additional information provided. You’ve ignored that it’s a man-made water feature in a garden of a 19th century national monument, just to make a point. I doubt about exercise, but ever heard about exploring similar areas? Maybe glance at the pond or even observe the marble handmade statue there (yes, the one that you wrote about before). Maybe it’s also a cool place to hang out with others? Maybe it meets the social bit, too?
Please, to avoid escalation, can you please refrain for commenting in this thread? Thank you.
I have had things rejected by the automatic system. We nicknamed it Emily because it sounds like ML which is short for Machine Learning. This is not an area I know much about, but it is doing more than recognise it as a pond. It is learns that this is the sort of feature that is likely to end up being accepted or rejected. As I understand it is always learning, and as with anything learning it can make mistakes. At the moment we know it rejects. I am not seeing as many false rejections appear here as we did when it was first introduced so it has been learning.
I absolutely get the frustration of having a nomination rejected by a machine before human eyes can start to assess it. The one advantage I have seen is in the review there are less, substantially less of the poor nominations than I would see before Emily was being used. Personally I miss the very quick acceptances that Emily did, thats a whole other story but the positiviyt of that experience shifted my view of the role Emily could have.
The Wayfarer Team are using this technology - thats where we are. They do read and take on board comments made in this forum. The general points have been made. Thats as much as can be done here with regards to the use of Emily.
Points have also been made to try and help move this particular nomination forward. If you want to take time and try to improve this one then do take the suggestions on board or we can discuss the options further,
Still, I’m trying to get Niantic’s attention and potentially overule the ML decision, per the facts given. It’s not fair to wait or use an appeal for something the ML got completely wrong.
