Bee Maps: A strong contender to Mapillary!

Community
1

Maybe I am the last to realize (because usually I’m quite slow :snail::grin:) but there is a new strong crowd sourced mapping contender in town:

Currently, they mainly focus on motor vehicle accessible roads and use proprietary developed dedicated dashcam hardware for capturing. Looking at the dashcam specs, it is indeed some interesting piece of engineering! Bee Maps has apparently well understood the need for consumer grade mapping hardware, something perhaps many have expected to come from Mapillary or at least to partner with some established camera maker. Okay, their $ 500 dedicated dashcam maybe is not particularly a consumer priced consumer device but you the gist. Well, the latest GoPro is not much cheaper either. They started in 2019 and they are really quickly catching up to Mapillary, basically in all aspects, from data collection workflow (own hardware), over data analysis and human in the loop data verification to API access. They also seem to reward contributors with crypto :key::moneybag: currencies or the like, something I wish Mapillary would have developed too or at least have thought about in the past few years. The only thing they do not do is contribute to OpenStreetMap. They build a private data stash.

1 Like
2

Hm, as you noted its a rather different business model to Mapillary used to have. Unfortunately monetizing access to the private data stash, is likely a necessity generate a revenue to support a contributors reward program; and this is likely implies that contribute to OpenStreetMap would be incompatible with the business model.

That stated, the Bee dashcam also support storing the collected imagery to a USB-stick, so you could most likely use to to contribute to Bee and simultaneously contribute to Mapillary; but the later would follow the traditional dashcam workflow. (The dashcam firmware is open-source, so might be possible to hack it to automate contribution to both…) Either way, I don’t think the contributor reward will cover the additional cost for the required LTE subscription for a individual contributor/hobbyist like me; though for a company the man-hours cost a performing daily manual uploads can add up quickly, so for that use-case the automation really makes sense.

The Bee dashcam spec is interesting, but the 13MP @30fps (e.g. AR1335 4208x3120p px) main + 2x 800p resolution might not be that much of an effective improvement over a the new now typical 8MP @30fps aka “4K” 3840*2160p main + 2x 1080p (FHD) resolution in of-the-shelf dascams, e.g: similar to other high-end of-the-shelf product, e.g Blackvue DR770X Box-3CH DMS Pro (or non-LTE VIOFO A139 PRO 3CH ); so other then the LTE connectivity, the main HW differentiator seems to be replacing the usual rear&and interior cameras with stereoscopic camera(s). My guess is that they are attempting to us that as a low cost substitution for a Lidar 3D-location-sensor.

I think the custom FW is what makes it really interesting, as if one can reduce the framerate to something more useful for mapping at the current velocity, then there is the possibility to increase image quality while significantly decreasing the bit-rate of data stream to upload over LTE. (As it’s supposed to double as traditional dashcam, it don’t seems like there will be any option to reduce the framerate (for the .MP4 files stored on the optional USB-C-stick, e.g. a 10fps mode, to reduce the storage requirements… ) In addition that FW likely also integrates with the Beekeeper fleet management cloud “AI” SW to only upload “interesting” images/videos, thus potentially significantly further reducing the amount of data to be uploaded…)

1 Like
3

Hm, maybe there are some part in that “Bee AI” that could be retrofitted into the Mapillary tools without too much trouble ? Specifically:
1: The automatic discard of “duplicate” contributions, as
where there already are sufficient (A) recent images with (B) good quality in the Mapillary database.
2: The automatic discard of images in excluded zones for privacy and/or legal requirements*, i.e your garage or the parking lot at your office.

(At least in Sweden there are more and more a few places that legally cannot be photographed (“avbildningsförbud”) for various security reasons, so the ability to disable the camera while passing these is a necessity to avoid making the automation a significant legal liability for the drivers…)

4

Unfortunately monetizing access to the private data stash, is likely a necessity generate a revenue to support a contributors reward program …

I do not think so. YouTube, like many other portals, work without a paywall. And, I am sure a business model similar to YouTube’s is possible for crowd sourced mapping. Sure, all or as much as it takes is a critical mass of contributors and customers to participate in this model but it is possible. This is an area where I would like to see innovation happen from mapping companies because I do believe it is doable. I do not have a plan or business model in my pocket here to share because I am not an economist but I am pretty sure it is not just possible but also feasible, especially since the advent of crypto currencies. Building a working business model is actually building a fabric of business connections. For example, most YouTubers are unable to sustain their content creation work just from YouTube’s ads revenue alone either, so they run Patrion and shops etc. Hence, it probably won’t be a one size fits all solution either. But, if we would be able to reward contributors even pennies for every image (which was infeasible with the traditional banking system but definitely is with crypto currencies) then suddenly things become a matter of scale and people are going to innovate to increase this scale. You cannot scale 0 but you can scale something > 0. :wink:

… and this is likely implies that contribute to OpenStreetMap would be incompatible with the business model.

This is not true because both aspects are disconnected. Note that imagery contributors do not contribute to OpenStreetMap, they contribute to Mapillary or whatever mapping company. It is a function of the captured imagery’s license that enables building OpenStreetMap map data from it or not.

Either way, I don’t think the contributor reward will cover the additional cost for the required LTE subscription for a individual contributor/hobbyist like me

Maybe, I do not know. Much of this depends on many aspects of the larger economic environment you live in. Besides, I do not really get the need for LTE, except for convenience. Because especially fleet users can automate WiFi access at their depots or parking areas to effectively upload via a wired connection, which in most parts of the world is still cheaper, more stable, and has greater throughput than LTE, but :person_shrugging: what ever… if it makes some people’s work or life easier then why not. So, I would not make LTE a requirement to contribute, especially because many parts of the world, including the US, still have no cell coverage at all but do have wired (or satellite) internet access.

the main HW differentiator seems to be replacing the usual rear&and interior cameras with stereoscopic camera(s).

It is really funny that they went with stereoscopic cameras because you do not actually need them to get depth data in the mapping scenario. :face_with_hand_over_mouth: You can just capture a parallel sequence or use a previously captured sequence to photogrammetrically reconstruct depth data computationally which is what Mapillary has been doing for years already. And, any effective space resolution and point accuracy is going to be better than this of these measly 800p infrared time‑of‑flight cameras, which also do not work on glass and chromed surfaces, or at certain angles at the sun! :grin:

Anyhow, new street level crowd source mapping companies appearing on the market means that Mapillary has to step up to their game to stay ahead. Unfortunately, Mapillary has basically lost about 3 years of development time because of the FB takeover. This is my impression at least because Mapillary began stagnating already about one to two years before the takeover. 3 years is a lot of time in the tech industry. In this market, you have to have something others do not have and I can see any advantage Mapillary may have gained slowly fade away. I would especially really hate to see Mapillary loose contributors because some other company has figured out how to reward crowd contributors even pennies.

5

For me Bee Maps would be too much of a risk/redflag.
You are required to buy their 500$ device.

Sure, they give u some pennies back, but nobody knows for how long. But with 500$ there is a long time they can pay you till they are even.

I would trust a project more if it is open to any equipment, like mapillary is.

1 Like