Home PSA Set Registry Forum

Computer Grading?

Been a while since I’ve been on here!
years ago they tried Computer grading coins.

To say it did not work out, nor was it popular would be an extreme understatement
Only lasted a few months as I recall somewhere around 1986.

Is it possible things could of changed since then? Yes and NO.

As with the coins - Computers can’t pickup subtleties of the cardboard as well as
Humans can (in my humble opinion!) just like with the metal on a coin.

Just my 2 cents worth.

Tony
KalineFan RIP my hero

Comments

  • JoeBanzaiJoeBanzai Posts: 11,171 ✭✭✭✭✭

    Welcome back!

    Why does one have to replace the other?

    I think a combination of the two would improve the process. A machine can measure the card and check for centering much faster and more accurately that the human eye.

    The human eye is also a great way to "pickup subtleties of the cardboard" that a machine might miss.

    At $300.00 a card the machine should be affordable?

    2013,14 and 15 Certificate Award Winner Harmon Killebrew Master Set and Master Topps Set
  • ArmourPhilArmourPhil Posts: 97 ✭✭✭

    I couldn't disagree more strongly. For anyone watching a baseball game on TV this is like do you trust the umpire or the "box" that sees the ball as it crosses a certain x,y,z coordinate ?

    I work in electronics and the optical inspection equipment available today is so much better than the human eye that this is not worth discussing. They are capable of looking at things that are sub micron. ALSO they can be tuned to grade within a certain window (i.e accept flaws that are no bigger than X ) . I don't know what they were trying to use on coins in 1986, but todays electronic equipment (after being set for the parameters your looking for) can replace every one of PSAs graders and from the quality of some of the work I have seen I say "don't let the door hit you in the ass on the way out"

    I especially love their PSA excuse catch phrase "eye appeal" which means "I don't care what you think we are not changing the grade"

    These kind of inspection robotics also dont get tired. Do the math on how many cards they are grading and how many seconds they have to come to a grade for each card. Do a few hundred in a row and tell me that doesn't affect your judgement ...robots don't get tired.

    Anyway if we all want the most consistent grading possible and really want to be sure that no one is getting favored treatment turn it over to the robots. Thats my vote !

  • JoeBanzaiJoeBanzai Posts: 11,171 ✭✭✭✭✭

    @ArmourPhil said:
    I couldn't disagree more strongly. For anyone watching a baseball game on TV this is like do you trust the umpire or the "box" that sees the ball as it crosses a certain x,y,z coordinate ?

    I work in electronics and the optical inspection equipment available today is so much better than the human eye that this is not worth discussing. They are capable of looking at things that are sub micron. ALSO they can be tuned to grade within a certain window (i.e accept flaws that are no bigger than X ) . I don't know what they were trying to use on coins in 1986, but todays electronic equipment (after being set for the parameters your looking for) can replace every one of PSAs graders and from the quality of some of the work I have seen I say "don't let the door hit you in the ass on the way out"

    I especially love their PSA excuse catch phrase "eye appeal" which means "I don't care what you think we are not changing the grade"

    These kind of inspection robotics also dont get tired. Do the math on how many cards they are grading and how many seconds they have to come to a grade for each card. Do a few hundred in a row and tell me that doesn't affect your judgement ...robots don't get tired.

    Anyway if we all want the most consistent grading possible and really want to be sure that no one is getting favored treatment turn it over to the robots. Thats my vote !

    It's nice to see someone who has an understanding of optical inspection equipment.

    I have been pondering this for quite some time. I have been out of the business for 20+ years, but I am curious how the machine would detect print dots.

    I am assuming every single card would have to be individually programmed for the machine to catch every flaw. Then just what do you do with an "obvious" print blob right on the players face as opposed to a "minor" one in the background?

    Are the machines advanced enough to "figure out" flaws now?

    I like the idea of humans having part of the inspection process, but not nearly as much as now.

    2013,14 and 15 Certificate Award Winner Harmon Killebrew Master Set and Master Topps Set
  • ArmourPhilArmourPhil Posts: 97 ✭✭✭

    Joe - optical inspection would not "figure out" flaws, but rather would compare the card being graded to a known species. That is certainly a lot of standardization , but I'm sure they would enter into this step by step.

    To me one f the harder issues will be how to remove the cards from the holders they were shipped in and insert them into a carrier that can take them through the optical inspection process without touching them. How many of our cards are dinged taking them out of our shipping sleeves now and we are never told about them ?? This may eventually require a new type sleeve that will be required to send you card into PSA. ...AND... they could return those with my order and not make me buy more of them.

    As an example lets look at a 1971T card.

    1. Coarse inspection of front and back of card identifies it as a 1971 Topps Ernie Banks #525. Computer confirms that this is the same as listing of the electronic customer order form.
    2. Coarse inspection measures incoming card for x and y dimensions and flags any cards that are less than a certain size.
    3. front and back inspection determines centering of the card matching given input parameters for what is acceptable.
    4. same inspection determines sharpness of the corners, front and back
    5. same inspection determines and chipping on the notorious back edges of the 1971T
    6. compare image to front and back image of known card . Flag flaws that are larger than a given size

    the above would be easy. I'm not sure how to set such a tool to see surface creases but I'm sure there is a way. If the eye can see it the tool can see it. The Optical Inspect tool need to be told "if you see this it means this" just like a child.

    Determining how to convert the inspection into a grade is another matter. We are told that every card now sees 3 graders to ensure consistency. Do we really believe that? How many of us have popped out a card from its case and sent it back in to get a higher grade. Nearly all of us I'm sure. I once popped out a Bazooka panel that received a PSA 4 and sent it back in to receive a PSA 7. You want me to believe 3 guys initially thought it was a 4 and the next 3 thought it was a 7 ?? I call BS.

    The computer, once the instructions are set will be consistent and show no favoritism to those with mega orders.
    This will no doubt take time and money, but it can be done and the result will be better than what we have now.

  • bobsbbcardsbobsbbcards Posts: 3,254 ✭✭✭

    @JoeBanzai said:
    I am assuming every single card would have to be individually programmed for the machine to catch every flaw. Then just what do you do with an "obvious" print blob right on the players face as opposed to a "minor" one in the background?

    It feels too time consuming to have a "reference" card for every one of the godzillions of cards out there, at least in the short term. More likely the AI wouldn't be able to tell the difference between an ink blob and the intentional ink on the card. That said, humans are uniquely suited to the "Hey, that looks out of place!" type of stuff, so maybe a quick eyeball test would work for print defects.

    Corner wear, scratches, creases, chipping, anything that messes with the cardboard itself would be easily detected by a computer ("Smooth 2-1/2" x 3-1/2" cardboard rectangle good.....beep.....boop").

Sign In or Register to comment.