Computerized Grading?
jrinck
Posts: 1,321 ✭✭
Other than for eye appeal, I don't see any reason why computerized card grading can't be a beneficial tool for the hobby. Almost all aspects of a card that can be quantified can be measured by a computerized system, for example:
CENTERING: If the card is identified as, say, a 1954 Topps card, the computer will instantly know all the caveats surrounding centering for this particular issue. A quick scan and fine-tuned pixel analysis can give a pretty darn good measurement of centering--much more accurate than a human can.
CORNERS & EDGES: The computer can perform a 3D scan and do a pixel analysis for all three dimensions, being able to precisely measure imperfections in either axis. Chipping will be most prevalent in the x-y dimension, and fraying will be more prevalent in the z-direction (for you fellow nerds). Also, if the card didn't measure up to normal dimensions, it could indicate trimming--which would also be detected by the 3D scan.
SURFACE: Something more than a 2D scan would be required, but scratches and creases could easily be identified. A more detailed analysis would measure gloss (if applicable to the issue). Printing snow and other imperfections would also be identified. Perhaps some sort of low power laser would be involved, especially in detecting recoloring.
Obviously the upfront cost for such a system would be great, but if this type of system proved to become the standard, then the cost to grade each card would be low due to volumes of submissions. Plus, you wouldn't have to pay it benefits, worry about bias, or ponder whether or not it was having a bad day on the day it graded your cards. But as with all computerized systems, a human would have to be involved simply for visual corroboration. The amount of humans necessary would be low, however, and they wouldn't have to be as highly trained or as expensive as the current grading system.
I know I'm not the first to think of this type of idea, and was actually wondering if anyone has ever taken it past the concept stage and either proven or disproven it's feasibility.
CENTERING: If the card is identified as, say, a 1954 Topps card, the computer will instantly know all the caveats surrounding centering for this particular issue. A quick scan and fine-tuned pixel analysis can give a pretty darn good measurement of centering--much more accurate than a human can.
CORNERS & EDGES: The computer can perform a 3D scan and do a pixel analysis for all three dimensions, being able to precisely measure imperfections in either axis. Chipping will be most prevalent in the x-y dimension, and fraying will be more prevalent in the z-direction (for you fellow nerds). Also, if the card didn't measure up to normal dimensions, it could indicate trimming--which would also be detected by the 3D scan.
SURFACE: Something more than a 2D scan would be required, but scratches and creases could easily be identified. A more detailed analysis would measure gloss (if applicable to the issue). Printing snow and other imperfections would also be identified. Perhaps some sort of low power laser would be involved, especially in detecting recoloring.
Obviously the upfront cost for such a system would be great, but if this type of system proved to become the standard, then the cost to grade each card would be low due to volumes of submissions. Plus, you wouldn't have to pay it benefits, worry about bias, or ponder whether or not it was having a bad day on the day it graded your cards. But as with all computerized systems, a human would have to be involved simply for visual corroboration. The amount of humans necessary would be low, however, and they wouldn't have to be as highly trained or as expensive as the current grading system.
I know I'm not the first to think of this type of idea, and was actually wondering if anyone has ever taken it past the concept stage and either proven or disproven it's feasibility.
0
Comments
My 1952 Topps Baseball Set
AAAARRRGGGHHH!!!
My eBay Store
BigCrumbs! I made over $250 last year!
PSA has graded 750+ cards for me and I can only think of one or two that dissappointed me, and thats only because of what I paid for them. There are people out there that have submitted infinitly more and continue to do so.
I like my Ryan's in their PSA holders and for me the Set Registry has been a source of fun and competition for the last two years.
Scott Jeanblanc
jeanblanc@iconnect.net
Ebay UserId : sjeanblanc
-----------------------------------------------
Collecting Nolan Ryan cards (68-94)
Well said... I agree 99.7% PSA does the job for me too... I'm in no rush to have my cards scanned like produce!
-Mike
I agree, startup costs would be enormous, but they would be amortized down over time--just like any new product.
As for throughput, if you limit the amount of mechanical contact with the card, the throughput could be quite high versus humans. If a human can give an accurate grade in 30 seconds, I suspect the computer can be more accurate in only two or three.
<< <i>And the really cool part would be when it hits the scanner and it jams and makes your card look like a bad grocery store receipt! >O
>>
That was sweet!
If such a technology is created for grading, then a technology that reproduces exact copies of the cards can't be far behind.
Jurassic Park, Dolly the Sheep...look out for the three-headed Wagner
Don't kid yourself........I have a 20x-40x microscope and while helpful, I still get 8's when 9's should be the norm. As long a humans are involved, nothing is for sure in the grading business.
BOTR
<< <i>I just got a hand held 100x Microscope, Real Cheap on Ebay. >>
Is there a particular one that you recommend? I bought a 10x from these posts and was dissapointed.
Collector
Topps 58,59,60,61,62,63,64 Sets
Fleer 60, 61-62 Sets
<< <i>Okay...it all sounds good to me...so where is the ultaviolet light that PSA uses to see flaws that the "eye" and I'm sure the scanner will not see...I can show you a card that you will not be able to see the flaw...yet under ultraviolet light it's there...how is this going to be addressed? >>
If the human eye can see it, the electronic eye can see it, too. You just have to tell the computer what types of things to look for.
I'm no expert on physics, but I would guess that in the scanning process the computer/laser would detect variations on the surface that could indicate recoloring, for example. The light from the laser would refract a bit different over, say, black ink used to recolor a '71 Topps. Same thing for white borders. Same thing for any type of surface.
And keep in mind that a human is involved in this process. The computer will spit out a grade, indicating the reasons why. A reviewer would quickly look it over and determine yea or nay. The intent would be that the human would always say yea.
All in all the computer process would be quicker than the 100% human process, and more accurate.
<< <i>first fully computerized collectible card grading program and our engineers have gone to great measures to bring you the very best quality encasement available today. Currently our facilities have the ability to grade up to 72,000 cards per month >>
CTA Grading wants you to believe somehow they have employed computer technology to 'assist' in the grading process but the best I can see is that they are using a 'calculator' (that's a computer?) to come up with some fancy equation to arrive at centering e.g.
Ahhhhh....one small step for mankind......what about cards that grade themselves! Make a note to Topps.....the card of the future......
Mike
Great concept - but it can't measure things like:
a) focus
b) eye-appeal
c) differentiate between background graphics and print spots
d) trimmed/altered cards (trimmed does not necessarily mean short, remember....)
etc.
We will see what happens!
<< <i>If the human eye can see it, the electronic eye can see it, too >>
One of my points is that there are thinks that the eye "cannot" see...yet PSA's ultra-violet light does see...I'm not sure a scanner will be able to catch these flaws...which for my part is not a problem...my problem is why is it a flaw if I can't see it unless I'm using "magic" light...in some cases I'd be more lenient while in others I'd be more strict...
Lenience for things my "human" eye cannot see...stricter for centering...I'd probably have more 10's and 9's...however those that finally arrived in those capsules...would be centered...pretty close to 50/50...
I'm a centering "freak"...
Collector
Topps 58,59,60,61,62,63,64 Sets
Fleer 60, 61-62 Sets
<< <i>...yet PSA's ultra-violet light does see...I'm not sure a scanner will be able to catch these flaws...which for my part is not a problem...my problem is why is it a flaw if I can't see it unless I'm using "magic" light...in some cases I'd be more lenient while in others I'd be more strict... >>
Henri
If memory serves me, I think they use a 'black light' which emits in the uv spectrum (eye can't see) to fluoresce a card - so if someone has used a touch of white-out on a border or certain coloring to enhance a card, it would fluoresce at a different rate than the normal ink on a card and SHOW UP for detection. Thus, if a card were rebuilt with rice paper, again I would guess the black light would pick this up. If a card were 'cleaned' - the phosphors in the cleaning agent would show up white on the card also. This is stuff way out of my understanding - does anyone have a source for this - a good article on the 'alteration' of sports cards. I think they can use the light also to detect different types of ink - so ink produced in Babe Ruth's era would probably fluoresce differently than ink produced today. Also, I do know that some paints today contain phosphors not in older paints - detection of forged paintings and again if a card were 'touched up' - the phosphors make the card 'glow' in the dark.
Just some ideas - Mike
I predicted that the distant future of grading would be something similar as jrinck described. It would significantly reduce biased (not eliminate, that would be impossible) grading.
However, it won't happen anytime soon.
I can't seem to get the site to come up so maybe it's out of biz. Anyway OGS or Online Grading Service was trying to get customers to submit scans and they would assign a computerized grade. Scans may be altered fairly easy, so that may be one of many reasons they might not be in operation. The CTA concept is similar, but uses the actual card and scans or image produces itself.
No grading service or method will be perfect, subjective opinion is alaways a part, no matter how small, of any final grade. The way PSA does its operations seems to be fine and quite acceptable to most. There is always room for improvement, but perhaps instead of looking for earth shaking innovations, more reliance on consistent and proven methods would be best to re-enforce our hobby and the graded card concept which is still realativly new.
In case anyone wants to try, the website for the OGS company is; ogscard.com
<< <i>Great concept - but it can't measure things like:
a) focus
b) eye-appeal
c) differentiate between background graphics and print spots
d) trimmed/altered cards (trimmed does not necessarily mean short, remember....) >>
I believe that computer grading could do all of the above. As far a focus, a scan of the card could be used to evaluate the alignment of the four color half-tone and the bleed and size of the dot to determine how well focused the card is. Printing spots, defects in the stock ect. could also be detected. You would need to have a data base of a scanned card to compare the card that is getting graded. If the two don’t mach up by a certain % then there is a print defect. Certain attributes give a card eye appeal, centering, focus, color saturation, whiteness of boarders ect... all which a computer could recognize. The more cards of on particular card that would go through the system, thus adding to a database of that card, the more accurate the grading of that card would become.
-Mike