I Wonder?

Here is the scenario.

They have an app for a smart phone to listen to a song and spit back the Title and Artist. Or go to a store scan a UPC and instantly get prices from everywhere else for the price on that item.

I have been to library book sales that people go around with hand held scanners / smartphones looking for books. If they find a hit to some sort of database they the purchase the book cheaply and sell at a profit. There is a lot of things that can be done with computer software here is my wonder.

What would it do for stamp collecting if you could go through your collection electronically and then correctly catalog your collection?

Or for the same as my above book purveyors. I wonder if it can't done for stamps as well by using some sort of stamp recognition software. A person could theoretically scan a picture, say off of ebay, and match it's bid price to a data base of selling prices. If the difference is great enough electronically purchase and turn a profit on a stamp, or even a group of stamps.

For stamp collecting a lot of stuff can be computerized but would we want to computerize it. At what point would the hobby still be called philately? Would we giving up the slow collection of understanding or knowledge on the hobby for expediency in processing the collection? Just some thoughts and ponders also what others think.

Well, off the top of my head...

Books have a UNIQUE ISBN number that identifies them; stamps don't.
Books have two grades: new and used. Stamps don't.

I think "philately" (the study of stamps) will always be with us, as stamps have too many variables to just be marked as merely an ID number (Scott or otherwise). Some have unique features (i.e. watermarks or printing anomalies or grades of centering or condition issues such as gum disturbances or perf pulls or whatever), plus add to that special postmarks or precancels or perfins or any number of other things that some collectors will value higher or lower than others.

Maybe one day Album makers will have barcodes printed for each space for a stamp in an album that will coordinate with a catalog app on an iPhone or whatever, but it is still up to the individual collector to study and identify the details that make one stamp more special than another.

Actually "philately" is the study of stamps, mail services, covers and postal history in general.

KirkS,

True to a point. I guess stamps are like diamonds where there may be a different perception on each individual piece. But If there was software available to look at an individual or even a pile of stamps. And correctly ID it down to a Country, variety and condition level all automatically would that still be called stamp collecting? I guess people don't need to know how a car is made to drive one or even how one is put together. But my question in a different way is this, would it still be stamp collecting if one doesn't know or care what a perforation is as long as a program does the tedious research we all have done through catalogs and observation to get to a conclusion on a stamp.

Effects on philately aside, I sure would like to have a data base that would ID (provide Sc #, date of issue & country) a scanned stamp, say from a mixture. There are stamps I have spent an hour or more trying to find in a Scott catalog, frequently without success. The next best thing is to scan it and put it up on Stamp Community Forum. That always provides quick identification.

Don

But therein lies the rub, DonSellos. Showing the Scott # would do no good for the majority of the philatelic world outside North America. Until there is a single standard numbering system for stamps (a stamp version of ISBN if you will) that all collectors and dealers around the world will agree upon, it makes little sense developing an app, unless you want it to be able to provide the catalog numbers for Scott, Gibbons, Michel, and Yvert et Tellier, plus any specialized catalogs that focus on the country the stamp in question is from. That could be done, but would require the Catalog publishers to give consent to use their numbers, which at minimum tends to be an expensive proposition.

The heart of the issue is that Philately remains in many ways stuck within national boundaries in an age where such national boundaries are becoming more of a hindrance than a benefit in regard commercial transactions and the sharing of knowledge.

The lookup for a catalog number is not a challenge (beyond licensing issues) but as wt1 points out the real obstacle is the recognition software. A mint stamp is pretty straight forward but once you get into used stamps with cancels recognizing the stamp presents some real headaches. An algorithm can handle removing some cancellations but keep in mind the diversity in size, shape and color. I've seen some software that get it 'right' about 66% of the time on used stamps but I simply do not see a way to improve it much beyond that. Maybe with a lot of work it might be dialed in to the 75% range but I doubt this would be good enough for most users. They will want the recognition to be north of 97%.
Don

A stamp recognition tool could be done with the technology that exists today, at least for stamps where paper types, watermarks and other (less visible) details aren't important. The main thing holding it back would be cost. There's probably not a big enough market for it to justify the expense of developing the app, generating and maintaining a huge database of images, etc.

I would agree that the market would be small for such an endeavor. And the price would have to be justified even if it did come to market. In my town they are using computer recognition to find bum parts in the manufacture of fasteners. Not a lot of data to check but very accurate in finding the bad item.

if the computer does all the sorting and cataloging of stamps according to the catalog, then the collector will not be able to cope up his free time. Stamp collecting starts as a hobby at an early age, then it catapults in to a passion, then sometimes you over invest in lots, then you could end up in stagnation of stock or would lead to stress.

Hence Philately could be helped by computer algorithms but the human mind is more powerful than the computer's hard disk and RAM.

Just my views as a collector, but the initiative is good on Vaccum Man's View point. I agree to disagree.

warm wishes.

It's a novel idea but it would be a mammoth task and considering all the varities and secret marks it would probably be more of a hassle than it would be worth, Still it's an interesting idea nonetheless!

The question boils down to…
Would you buy a product like this if the recognition on used stamps was less than 75%? 80%?
There is no way on earth that recognition software would ever be able to be written to 'uncover' a killer cancel, it would struggle (at best) to separate any black cancel from a black and white engraved stamp. Think about how many stamps you have seen where the cancel has covered an important characteristic of the stamp needed for identification. At best the software would only be able to give you a choice of possible candidates on some stamps. There is existing software that does do stamp recognition and frankly it sucks. Even using reverse image lookups from major providers like Google and TinEye fail miserably when using them against an image of a used, cancelled stamp.
When people fork over good money they are not going to be happy when it 'doesn't work' even some of the time. If this is like every other type of recognition software ever marketed it would have to be consistently above 98-99% for users to consider it for purchase.
Don

My biggest struggle with the hobby is that it takes up a bit of time to go though and correctly ID a group of stamps. Even more so for the country's that I am unfamiliar with. Even at 50% recognition would it save time? There would probably be some sort of time factor in getting the visuals set up so that a system could actually use the information to make a good determination.

Back to my original question (or questions) of, If there was such a system out there would stamp collecting cease being a hobby? I don't think so just because of the many variables involved. There would still be a need for someone to expertize the stamp. But it would make it easier and faster to find that stamp that begs to be expertized.

Insurance / accounting - wise I think it would be a boon, and allow for special modifying identification of special stamps by certificate or by considered advice.



I think the hunt and personal comparison of hidden features would almost preclude or rule out the use for this idea.



Collectors collect, and hunting, seeking, valuing, sorting and appreciating is what is to be gained as well as sought after.

vacuum man...I have been thinking about that for a month trying to figure out how/what software better attacks the problem of stamp recognition as in face recognition..I have been investigating approximately 17 face recognition programs to see if they can be adopted to stamps..There are some problems that arise with this technique..

1 - Stamps change colour over time.
2 - Shades are not recognizable that easy.
3 - Cancels distort the finished product.
4 - Who will take care of updates/additions
5 - etc...etc...

I am still looking at this software based Unitrade stamp identification process....It is beginning to look harder that I first anticipated.