Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

oo #20

Closed
justgeek opened this issue Oct 18, 2014 · 10 comments
Closed

oo #20

justgeek opened this issue Oct 18, 2014 · 10 comments
Assignees

Comments

@justgeek
Copy link

No description provided.

@justgeek justgeek changed the title This is crap oo Oct 18, 2014
@foo123
Copy link
Owner

foo123 commented Oct 18, 2014

i'm not sure what is the issue here, it is closed. In any case even if it is for criticism, its ok

@justgeek
Copy link
Author

the issue is that your code is not working on any image but the one attached with source !!

@foo123
Copy link
Owner

foo123 commented Oct 18, 2014

yes this can happen, i say it is not the code, it is the type (and difficulty) of the application.

In general image recognition is a very difficult task. Of course if the code has issues, it is good to know.

@foo123
Copy link
Owner

foo123 commented Oct 18, 2014

Try to play around with the detector paramaters and see similar issues #19 and #17

@foo123 foo123 self-assigned this Oct 18, 2014
@justgeek
Copy link
Author

Actually there are much better libraries out there , the problem is im trying to detect human neck , never found a cascade for that :(

@foo123
Copy link
Owner

foo123 commented Oct 18, 2014

You can train a cascade for anything you want, there is a link in the readme of this repo, gather any image samples that cover the feature you need and train one.

Well there are of course other js libs like liuliu/ccv (the 1st for face detection in pure js), js-objectdetect and jsFeat which cover similar functionality (some with additional features as well).

You can also try another lib of mine filter.js for general image/video processing where you can add custom plugins (e.g HAAr.js can be added as plugin)

Hope this helps

@justgeek
Copy link
Author

thanks alot for your concern , but would you explain how can i use haar.js with your filter.js ?

@foo123
Copy link
Owner

foo123 commented Oct 19, 2014

One can add it a s a plugin in filter.js, include the haar.js file and filter js and make a custom plugin, which can be used as a filtyer with other filters.

haar.js does not support filter.js plugins by default but one can make almost any other library or code into a plugin for filter.js (if it so desired).

FILTER.Image can accept any url/image/canvas/video element and represent it as a filter image which can be processed.

FILTER.js is a work in progress and aims to support multiple ways to process images/video in pure javascript (inluding main thread js, parallel threads, webgl, svg and css filters)

Most of these features are already usable (main thread, paralle threads and webgl in progress).

haar.js was made to function standalone without any dependencies (plus optionaly support parallel computations with parallel.js)

So no extra code is added, however f this is a popular request i can customize the haar code to be used more easily as a filter plugin

@justgeek
Copy link
Author

but this has nothing with training custom objects right ?

@foo123
Copy link
Owner

foo123 commented Oct 19, 2014

No, training cascades is another thing. If a cascade for a feature does not exist, one can as well train one.

The haar code (either standalone or as a plugin whatever) will use cascade for detecting a feature. The creation of the cascade (i.e training) is another thing, one will need the openCV package for this. See the link in the readme file.

Plus one should be aware that the quality of detection depends on the quality of the cascade and the quality of the cascade depends on the number and quality of samples used to train the cascade.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

No branches or pull requests

2 participants