RE:RE:RE:RE:$0..27 So I am stupid !? Wow, that's a really strong argument. I am impressed.
So would you like to share with the audience, what you know and not spectulate about ?
Maybe You could answer a few simple questions in the first place:
As we heard, Slyce's visual search software is implemented in apps of home depot, toys-r-us app, neiman marcus and some others. Generally the user of the app takes a picture of an item and sends the piture to a server.
As I understand from your impressive comment, You believe, that this picture pops up on a workplace of one of many operators, who select the right article or a few similar and send that back to the user app.
- How much time do you think a single operator needs,
when a picture of a user of the app pops up
at his work-station an he then searches the right article or
a set of similar articles and send it back to the user ?
Is 5 seconds average a reasonable assumption ?
... to see the picture, translate the picture
into words (skirt, red, short), punch the words
in the keyboard and select from the list of articles ...
Or is 10 seconds more reasonable ?
- How many people do You need, in order to do - let's say only - 1.000
of such searches in an hour ?
If the operators need only a sporty 5 seconds average, one of them can "solve" 3600/5 = 720 searches, if he needs 10 second he still can solve 360.
Ok, so statistically 2 operators could do, without having breakfast, lunch, go to the toilet and such and if the searches come in one after the other in a constant flow. So what, if 30 searches come in in a single second. Let the last of them wait 150 seconds to get an answer ? Hm, obviously, 2 operators can't do that.
So, how long is the longest wait for a result on either one of the apps ? 5 seconds ? 10 seconds ? Let's assume 10 seconds - which is so long, that I would never use the app any more - then you needed already 150/10 = 15 operators to deal with such a peak request time.
Do you see the picture ? How about 10.000 searches per hour ? 150 operators ? 20.000 searches ? 300 operators ?
If that is the real process like you think, then WHERE IS THE BUSINESS MODELL ?
I am in the software business since 1977, run my own company since 25 years. I am closely following the developements of pattern recognition for a very long time and I am not claiming, that I am the top specialist in the area but I would say I have a certain insight.
If you want to know, why I explained the technology like I did in the last post, then i suggest to read:
Selective Search for Object Recognition
Here is a link: https://link.springer.com/article/10.1007/s11263-013-0620-5
HvT