API Harvest: How to know whether an image is from a 360 degree camera?

Hey Mapillary folks,

When harvesting via API, is there a way to ascertain from the image metadata whether the image (or sequence) is taken from a 360 degree camera? The information should ideally lend itself to be stored as part of the harvest metadata, just like key, ca, coordinates attribute information. Possible reason for this would be to, for example, harvest only 360 degree images, or all but 360 degree images.

Regards,

Momo

Hello Momo!

In the API response, each image contains a property called “pano” which is a boolean true/false.

You cannot filter the API to return only pano=true or pano=false, but after receiving the response you can parse the features to determine if they are 360 degrees. I know this means more client side processing, but it’s the only way to achieve this currently.

1 Like

On the web interface it’s possible to filter by 360, so what is the limitation of the API to not provide the same feature? If it’s just time and resources, that’s ok, but the functionality is there on system, somewhere, and available on the web, so the worst is already done, isn’t it?

The limitation is that the database doesn’t automatically code each photo as 360 or not, so applying the filter can’t then query the whole database that way. Instead, after the query is complete, the photos are labeled as 360 or not. The short story is that the earliest versions of our database didn’t plan for this to be searchable, and the website later got a hasty fix to allow filtering in it, but we have not prioritized a reprocessing of the database to make this filterable generally in the API.

I hope to see this function fixed in the future, as I think it has a lot of uses going off of what Momo mentioned. Our database and API experts are well aware of the issue, so right now it’s just a matter of priority!

2 Likes

awesome thanks

That is a bit the trouble. We usually don’t know what Mapillary is aware of.
And we come with proposals.
Has Mapillary thought about filtering OUT people ? It would be useful when surveying highway pictures in my region.

Ciao a tutti,

thanks for your input and clarifications. I was aware of the “pano” statement, but was not sure whether pano necessarily means 360. In your experience, are pano=true images also always 360?

MoMo

They are not always 360degrees. I had previously uploaded some panorma shots taken on my iphone (approx 270deg horizontal field of view) and it parsed it as a full 360 panorama.
For example:
https://www.mapillary.com/map/im/2EoXURgYnLrrao35eRvHbg

1 Like

Ok, great input, everyone! I shall return when I have news on how to work around this. In the meantime, should anyone develop something to ascertain this 360 information easily, please do share! Peace from the East! M

1 Like