search

ericblade / quagga2

An advanced barcode-scanner written in Javascript and TypeScript - Continuation from https://github.com/serratus/quaggajs
MIT License
758 stars 85 forks source link

node-test example not working well #314

Closed pguardiario closed 3 years ago

pguardiario commented 3 years ago

When I run the node-test script I see it works for test/fixtures/code_128/image-001.jpg, but it only gets 3 out of the 10 in that folder, and it crashes on some of them. Is this expected behavior or am I doing something wrong?

github-actions[bot] commented 3 years ago

Thank you for filing an issue! Please be patient. :-)

ericblade commented 3 years ago

I'm so glad someone is running tests! :-D I've spent a LOT of effort on upgrading the test suite from what we had originally when forked! :-D

It is known that several tests fail, it is my hope that we will maybe perhaps have some people working on improving image recognition, and getting any of the tests that are currently failing without also breaking any of the ones that are working would be a great place to start :)

All of the tests that are failing now were also failing when I forked the code, although 2 tests that were failing before have started passing since then, so there's been some improvement (i believe they were in the code39 reader, so not too helpful for a lot of us, i think most people are using ean/upc/code128).

Right now when any changes are made, I compare the test results to previous test runs, and as long as the number of failing tests doesn't go UP then i'm happy.

Does that answer your question ?

(it might be possible that changing some of the reader / locator settings may improve test results, that would be a relatively quick thing to adjust, but i'm hoping also that code improvements will help)

pguardiario commented 3 years ago

Kind of, I was wondering if those images were being recognized correctly elsewhere and just not by the node-test.js script.

On Wed, Jan 27, 2021 at 3:59 PM Eric Blade notifications@github.com wrote:

I'm so glad someone is running tests! :-D I've spent a LOT of effort on upgrading the test suite from what we had originally when forked! :-D

It is known that several tests fail, it is my hope that we will maybe perhaps have some people working on improving image recognition, and getting any of the tests that are currently failing without also breaking any of the ones that are working would be a great place to start :)

All of the tests that are failing now were also failing when I forked the code, although 2 tests that were failing before have started passing since then, so there's been some improvement (i believe they were in the code39 reader, so not too helpful for a lot of us, i think most people are using ean/upc/code128).

Right now when any changes are made, I compare the test results to previous test runs, and as long as the number of failing tests doesn't go UP then i'm happy.

Does that answer your question ?

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/ericblade/quagga2/issues/314#issuecomment-768107471, or unsubscribe https://github.com/notifications/unsubscribe-auth/AOASJVUZ4KFPH4F6PPE6ROTS37BXFANCNFSM4WS7SCGA .

ericblade commented 3 years ago

i really haven't tried them in any other configurations. my assumption is that a few of them might be picked up with some adjustments to the configuration, but it wouldn't surprise me if there's not enough whitespace around them for several. i just haven't really concentrated on trying to improve the actual detection and decoding codes at all, i've spent the entire time since forking just modernizing the code and getting it to work reasonably with typescript.