For the bench photo, you don’t show the whole bench. My guess is that the ML model could not decipher those slats as anything significant. Try turning your camera sideways for a new photo more similar to the ones you linked.
For the park photo, there is no visual anchor. Something needs to be in the photo that will allow folks to identify that they have reached the Wayspot. A bench, a sign, something besides just a lot of grass, as in the photo example here: