Hivemapper - State of the Hive 4: Tokenomics of Bursts

[@davidvalenzuela8733]

Great stuff guys. Keep up the good work!

[@chats5957]

There is no point doing any AI at the edge. The HDC isn't connected directly to the internet so it cannot do any useful image/point cloud processing because it won't be comparing current imagery with previous captures. The only processing that can be done at the edge is privacy masking faces and licence plates. They claim that they're already doing that, but maybe they aren't? To create 3D models of roads and road assets, you need multiple data sources from different angles and much more processing power than a Pi cm4 with 2gb of ram. The whole AI at the edge announcement is just to create hype and attention in an effort to sell more cameras. The lifespan of a HDC is already subpar. Trying to milk more out of it to save server storage and processing costs is going to cost more in the long run if they cover failed hardware. So many aspects of this project just don't make any sense.

[@Future_Networks]

Thanks for the in depth take! Apparently the AI edge processing is driven primarily by customer preference, I.e. to prevent unblurred images from making it to the server, not to save on AWS fees. So it is exactly for the reason you stated - privacy masking faces etc. Although, lower costs is a nice side effect that also helps with scalability. I disagree that the current hardware isn’t up to the task of creating adequate road models from driving both ways down the same street - we’ve seen snippets of the map built by these cams and it’s impressive.

[@Maf12300]

What about the open road contest honey token air drop?