I’ve noticed that since the implementation of the Machine Learning that several architectural elements I’ve submitted have been rejected nearly immediately, within twenty-four hours. I’m talking specifically about elements like large stone carvings, bas-relief sculptural embellishments, and the like. For example this carving of a satyr head was rejected by the ML twice before going into reviewer voting.
Title, translated: Satyr Face
Description, translated to English:
This horned figure stares at pedestrians in Viale Campania. Whether he is a satyr, a devil or a plain old goat is up to the viewer. This face, however, remains enigmatic and carved in stone.
Supporting, translated: This unique sculpture sits above the threshold of a building. Located on a secure pavement with 24/7 access. It is located outside a law firm, with access via tram and bus. This sculpture is located near several parks and a museum. It is fully visible on Google Street View.
Not the best supporting statement, I’ll admit, but I thought it got the point across. Anyways, I’ve had similar ML rejections four times now and it seems unusual to me. I was wondering if anyone else has noticed anything similar, like perhaps the ML simply isn’t recognizing carvings and decorative facades as being acceptable? I’ve appealed my lastest rejection and hopefully either it’ll help the ML recognize that these can be great places to explore, since it keeps saying Wayfarer criteria as the rejection reason. So yeah I was mostly curious if anyone else has noticed anything similar? Or is there another category that the ML is consistently not getting correct? We know the ML sometimes gets confused when there’s too much nature in a photo…
