I'd like to preface this with the fact that I have next to zero experience about the topic of pricing/charging for such service's, so what I say might sound naïve, but…
Is it really that expensive, though? I mean I can see it being expensive if you use it inside a product you offer for free, but if it's a commercial product, I imagine the pricing isn't that much especially if you consider that using your own infrastructure for this type of machine learning and image scanning would be much much more expensive.
I have no insight into the actual company's thinking, but in general there's not necessarily any relationship between the cost to provide a feature of software and the amount that is charged to use it. It's usually more about willingness/ability to pay. If most of the people who need access to the raw data have a certain use case that means they're willing to pay for the expensive plan, that's a good reason to charge for it, assuming your goal is to make money as a business.
I'm a bit confused by how they priced some of these things, can anyone help me understand?
• Deep Neural Networks Module: $150k
Considering how important deep learning is for computer vision and image processing in general now, this is a very important. But will this be a completely new library for both training and inference? Why not use what's already out there, like Pytorch or ONNX? $150k pays for 1-3 developers for a year, depending on where they live.
• Accelerated image processing: $150k
I don't know what this means in practice, refactoring the code adding more SIMD code? OpenCL kernels? Again, the price tag seems steep, but maybe they will hire an expert for half a year?
• Improved support for fisheye camera calibration: [$100k + hardware ($20k approx) shared with multi-camera support]
I know OpenCV is often used with very expensive research cameras, but couldn't this development be done with a couple GoPros? If companies want to certify it for their esoteric cameras let them pay for the added expenses.
• $700k stretch goal, CI and build server.
Does this cost $100k?
Don't get me wrong, I like OpenCV, I've used it and will donate to this, but I don't get how these things can cost that much for an open source project. There's no price attached to the improved documentation and tutorials, those are important and take a lot of time. It's probably faster to write a canny edge detector than writing all the documentation and examples for it.
On a separate note, their $1M stretch goal mentions a future OpenCV cloud service. That seems like something that should be sponsored by one of the cloud providers, or handled by a separate company instead of a crowd funded open source library project.
As other posters have mentioned, the incument companies rebranding to Observability definitely are expensive, because they are charging in the same way as they do for logs and/or metrics: per entry and per unique dimension (metrics especially).
Honeycomb at least charges per event, which in this case means per span - however they don't charge per span attribute, and each span can be pretty large (100kb / 2000 attributes).
I run all my personal services in their free tier, which has plenty of capacity, and that's before I do any sampling.
A fair point and it is a cost associated to the experiment we're running.
If you have a sincere interest in the software, you can convert on one of our forms on the website and you'll hear from a sales person. You can get pricing relatively easily in that fashion.
There’s a market for that though. If I am running a startup to generate video meeting summaries, the price of the models might matter a lot, because I can only charge so much for this service. On the other hand, if I’m selling a tool to have AI look for discrepancies in mergers and acquisitions contracts, the difference between $1 and $5 is immaterial… I’d be happy to pay 5x more for software that is 10% better because the numbers are so low to begin with.
My point is that there’s plenty of room for high priced but only slightly better models.
That compute isn't free - it comes at the opportunity cost of being sold to a paying customer so it makes sense to price it internally even if it is "virtual"
Interesting - thanks. I have a friend who used to work at Coremetrics. I didn't realize it was so expensive. Maybe it wouldn't seem like so much if we weren't all seduced by the free Google stuff.
One that drives sales in the retail sector. “Do I have to be online in order to accept scans/payments/etc?”. If I’m a small business owner and want to scan my stock, I have to pay for the privilege, when my USB scanner works without internet and costs me $160. I really don’t think this will fly at the price/subscription they are selling it for. Especially when there’s cheaper/free options available. [1]
Anecdotal - for a large publicly traded financial services who wants to manage their book of record software (i.e. the software that holds the truth of "dear client you hold this many assets"), you're looking at $3-5M/year in license costs.
Depending on the business context of a ML solution for a company like the OP describes, you're likely looking at $500k+, but this is pure speculation and requires more information about the customer situation (e.g. is the ML solution going to save the company money or help them sell more).
Thanks for the figures. I suppose with expenses like that, they will be motivated to research methods of updating models which have already been trained.
I don't know about the price but you are not wrong. A lot of naysayers just don't know how to use it well enough in their workflow. This is definitely the future of the industry though.
I'm not sure I understand the value proposition. Supporting abandoned scanners is great, but $40 for essentially a driver seems rather steep. That's half the cost of a brand new scanner that will work with whatever software I want.
To be clear, this would be peanuts to my company. But it's super expensive for what it is especially because they have a free service that does the same thing right next to it.
Genuine curiosity, who is paying for that type of work? What is a typical use case? Where is the ROI? I'm not familiar with this world at all. I appreciate any insight!
Is it really that expensive, though? I mean I can see it being expensive if you use it inside a product you offer for free, but if it's a commercial product, I imagine the pricing isn't that much especially if you consider that using your own infrastructure for this type of machine learning and image scanning would be much much more expensive.
reply