Robotic graders? Best idea? Or stick with humans. Or 50/50?

[KingPokemon]

I got the answer while I was watching Smpratte’s new video on Machine Grading. All you lurkers, get in the game and watch it here: smpratte - Should PSA Use Machines To Grade Pokemon Cards? - YouTube

Are you sitting down? Here you go.

THE E4 POST-GRADING MACHINE

Send in your graded cards to e2d2 for scanning and the results, along with the cards serial number, will be posted here on the website.

Sounds like a million dollar idea to me…

[smpratte]

Haha if someone has the technology, and wants to start a new grading company, I’m a potential investor!

[Kuja]

Awesome video, Scott!

Count me in as well.

[saber]

@pfm

@KingPokemon

What would it cost from inception to permanent use for each machine? This is really what most of it comes down to. Like I said earlier, many minor surgical robots are being purchased all over the world for 2 Million each plus many additional costs. Well that’s way less than the initial finished product cost to come to market afte r&d etc. That investment made sense though cause thousands will be sold. Not true with a card grading machine.
So again, what would it cost total for that first one to come to market?

In terms of hardware the solution is not very demanding. Maybe put some money to a good scanner, say $2-3k. Probably need a place to store the raw data. $1.5k could get you a 10TB harddrive which should be sufficient. The cost comes in collecting data and developing the algorithm.

You’d have to get someone to scan the front/back of cards. You could pay them $12/hr and it would cost about $25k per year. The data could be used for other purposes such as expediting damage claims since you now have a high-quality scan of the card prior to grading.

After enough data collection, you’d need to hire a software engineer to develop the grading algorithm for you. That’ll run you something like $80-120k per year. I’d imagine to get a decent prototype up and running would take less than a year.

So probably something on the order of $100k. Since it’s a bit of a research project, you could probably apply for government research grants to help mitigate the cost.

I work at a University library and we have a very expensive scanner. It’s called Zeutschel and it runs for about 10K and that was probably discounted. It’s a very high tech (big) scanner. You might need a couple of them, who knows.
As for the algorithm, you’re going to need more than just one software engineer. (I asked my husband, he does stuff, with codes and numbers ). Probably might need a small team (maybe 3-5) of experienced coders and depends on where they live you’re looking at minimum 80K to 100k each, for years of experience and cost of living. Your raw data is probably going to be higher, and probably looking at having at least one IT person to keep an eye on it. You know servers are fussy sometimes.

I just wanted to say something about the numbers.

[pfm]

@saber

@pfm

In terms of hardware the solution is not very demanding. Maybe put some money to a good scanner, say $2-3k. Probably need a place to store the raw data. $1.5k could get you a 10TB harddrive which should be sufficient. The cost comes in collecting data and developing the algorithm.

You’d have to get someone to scan the front/back of cards. You could pay them $12/hr and it would cost about $25k per year. The data could be used for other purposes such as expediting damage claims since you now have a high-quality scan of the card prior to grading.

After enough data collection, you’d need to hire a software engineer to develop the grading algorithm for you. That’ll run you something like $80-120k per year. I’d imagine to get a decent prototype up and running would take less than a year.

So probably something on the order of $100k. Since it’s a bit of a research project, you could probably apply for government research grants to help mitigate the cost.

I work at a University library and we have a very expensive scanner. It’s called Zeutschel and it runs for about 10K and that was probably discounted. It’s a very high tech (big) scanner. You might need a couple of them, who knows.
As for the algorithm, you’re going to need more than just one software engineer. (I asked my husband, he does stuff, with codes and numbers ). Probably might need a small team (maybe 3-5) of experienced coders and depends on where they live you’re looking at minimum 80K to 100k each, for years of experience and cost of living. Your raw data is probably going to be higher, and probably looking at having at least one IT person to keep an eye on it. You know servers are fussy sometimes.

I just wanted to say something about the numbers.

The scanner doesn’t have to be sophisticated. For machine vision tasks, images are often scaled down and this process tends to make learning simpler and more generalizable. You can probably get away with a $200 scanner, but obviously it’s easier to reduce the resolution of images than to try to increase resolution so it’d be a good idea to invest a bit more in a high quality (and fast) scanner. The one you’re referring to might be a bit overkill since you wouldn’t need the resolution necessary for, say, scanning pages of ancient books.

You’re definitely right that you’d need more than one individual to do things such as maintaining the data. But PSA presumably already has a team of people who maintain the population report and build their search features/auctions prices realized features. They already got the infrastructure in place, just need an additional database for storing images. It might be ideal to get a couple people working on the machine learning algorithm project, depends on how much you want to prioritize this project. I’d argue you could probably get away with one just to see if you could make a proof-of-concept prototype and to minimize the loss should the project fail for any reason.

If you’re building a start-up company from scratch, you’d probably want 3-5 people as you mentioned and that should factor into the cost. But at PSA a lot of these resources already exist in the company based on the services they already provide.

[poken00b88]

I think your numbers are too low, but even if it were say $2-5M for PSA to develop this on a larger scale, it will pay for itself over time. They grade over 1M cards per year. It’s worth it because they have the volume, this isn’t like the people wanting them to make a special case for graded Pokemon packs that would cost an arm and leg to grade maybe a dozen of these things per week.

As you said they could also run some kind of beta version on the cheap and see if it’s feasible. They could test say 10,000 cards from different hobbies and see what kind of results a machine will come up with vs human grading.