mtschirs / js-objectdetect Goto Github PK

Can js-objectdect be used to detect simple geometric object in an image ? Can there be an example like http://stackoverflow.com/questions/11424002/how-to-detect-simple-geometric-shapes-using-opencv

Is there any tool / library to create new classifiers for nose

Why does the "jerks" of the eyes ,image occur even when the face is constant. Is this an issue to be fixed?

Hi!

Right now there are differences between master and gh-pages example. Not sure how many there are, but I was trying to use gh-pages branch (as submodule) [which is my own issue for doing it wrong] with code from example and it turned out, that master is featuring
https://github.com/mtschirs/js-objectdetect/blob/master/examples/example_image.htm#L11-L13 (objectdetect.nose.js), which is not available in gh-pages branch.

Maybe, it would make sense to have some sort of Travis CI setup to push everything from master to gh-pages?
Or, maybe, it's possible to set gh-pages as default branch for the project so that all up-to-date code is in gh-pages branch & live examples on actual github pages?

Hello there,

I trained a custom classifier using opencv_haartraining and convert the xml result using this, #7 (comment)

I test the unconverted xml using opencv sample facedeject and its working fine.
Do I missed something?

I also experimented changing the value of scaleMin and scaleFactor, still doesn't work.

Any help is much appreciated.

Thanks.

Hi.
Can you tell us the way you converted the xml classifiers from OPEN CV to your js fle format ? I tried the HAAR.js converter, but the generated js file is different from yours.
I actually need the nose classifiers.

Thanks ;)

Would you have any tracker from the mouth?!

thanks

I do not see any links to the jsdoc documentation in the source file as stated in the README. Does anyone know where it is?

Probably shouldn't have hidden the request to change to an MIT or Apache license as I'd love to reuse this in some of my own work (and contribute of course).

Based on only a cursory look, I think MIT I think is likely the best license to allow re-use.

Great library, and thanks in advance!

Sincerely,
//Adam

It'd be great it https://github.com/mtschirs/js-objectdetect#performance would include also an approximate size of the library as a WebApp would see it.

Hi, I'm trying to use fist and open hand classifiers. I'm getting a lot of non-sense finds. I start changing the scaleFactor parameter from 0.1 to 2 and I didn't get so much success finding a best scale factor. Some times it works fine but I got a lot of non sense findings.

I started using video stream directly. With much noise, I tried to migrate to find on a gray image instead.

I got crazy with so much signals with no connection with real hands.

I put red for fist hands and kind of orange for open hands. The face is recognized by jsfeat classifiers and it uses a green square do identify the face. The same method I used to plot face, I plot for hands.

I don't understand how much sensibility of camera or even the not blank background can break the things here. I got some nice finds too. But so much noise was find as hands. Have some one deal with something like that?

Hi, i'm currently started a library project https://github.com/inspirit/jsfeat
and while working on porting my own version from flash to js of haar detector i found your implementation. so i wonder if u agree to merge your version to jsfeat?
i already implemented faster canny, grayscale and gaussian methods so your version will be even faster and more responsive :).
let me know what u think.
Best.

I am after a valid example of your objectdetect.js to detect eyes in a face ROI. I have succeeded to detect one eye and integrate into the image example. I would like to detect eyes within the face being detected, can it be done?

this will detect the face and one eye, what do I nee to do to make it detect 2 eyes in a face region, can I make it work with many faces at the same time?

on openCV C++ I did manage, but in JS I am having hard time, can u help?

", { "css": { "border": "2px solid #00FF00", "position": "absolute", "left": ($("#face").offset().left + coords[0][0]) + "px", "top": ($("#face").offset().top + coords[0][1]) + "px", "width": coords[0][2] + "px", "height": coords[0][3] + "px" } }).appendTo("body");

        $("#face").objectdetect("all", {classifier: objectdetect.eye}, function(coords) {
            $("<div />", {
                "css": {
                    "border":   "2px solid #F00",
                    "position": "absolute",
                    "left":     ($("#face").offset().left + coords[0][0]) + "px",
                    "top":      ($("#face").offset().top  + coords[0][1]) + "px",
                    "width":    coords[0][2] + "px",
                    "height":   coords[0][3] + "px"
                }
            }).appendTo("body");
        });

    });
});

</script>

i can see files like :
objectdetect.mouth.js

where is this used and for what is it made can not see it being called

Your framework is very well built. I'm using the framework on both video and images and while single faces are ok...when using multiple faces or videos i get a lot of false positives.

is there a way to fine tune the detection ?

Hi

Running your demo with the glasses - and the same code locally in testing - I notice that the performance of the tracker is about 3-4 FPS in FireFox on Windows and OSX but easily manages 20+ FPS in FireFox on Linux. Also, the same code executing in Chrome and MS Edge browsers on all operating systems runs at 20+ FPS. I have tested on several machines with Windows 8, 10, and OSX.

There seems to be some fundamental issue with Firefox and the tracker detect() call.

I will dig into it a bit more but it'd be great if you could take a look into it as well. I will update the issue if I find anything and submit a pull request if I can fix it.

Hi!
It would be nice to add package.json file. Even if you don't publish this project to npm, people would be able to run npm install mtschirs/js-objectdetect which will enable easier usage with bundlers like browserify/webpack.
Right now you would get an error of following sort:

npm ERR! eisdir EISDIR: illegal operation on a directory, read
npm ERR! eisdir This is most likely not a problem with npm itself
npm ERR! eisdir and is related to npm not being able to find a package.json in
npm ERR! eisdir a package you are trying to install.
npm verb exit [ -21, true ]

variable "msDiff" calculated in smoother.js line 24 is always 0 (i assume it should store the time elapsed between two anim.frames?)

I cloned the repository to how the library works, and tried both of the examples and none of them work (console.log(faces.length) in the image with faces example returns 0).

Hi!! Really great library!!
I would like please to know how could i create a new classifier (model) for recognition like objectdetect.handopen.js. I would like to know is there any tip :P
Thank you for your great job!!!!

I am unable to get the objectdetect.handfist.js to work, Can you add an example on how to use this.

can i detect neck using your technique or can you provide me guide how can we do it.

i need to make application like add neckales to user neck

I'm liking your project, but I find your classifiers to be spotty with their accuracy. I would love to make some new classifiers, how would I go about this? Did you convert xml haarcascade files to js? Help would be much loved! Thanks

Tried to use the mouth or smile detector with this libary they just 99% of false postive.. firing all crazy randomaly places tried to limit so confidence>10 still not help.
than I tried to use the smile detector from opencv I tranform it to array and when tried to use it I get an error of

Uncaught TypeError: Cannot read property '7' of undefineddetect @ objectdetect.js:508detector.detect @ objectdetect.js:717play @ a.html:64

this the not working converted smile detector:
http://jsfiddle.net/hnjtL5zy/

I even tried to use 2 detectors so I detect a face and than give this to the smile detect:

//detector = face detect;
var coords = detector.detect(video, 1);
if (coords[0]) {
coord = coords[0];
var coords2 = smile.detect(video, 1,1,coord);
etc...

still not working give me a lot of false positive randomaly in the camera not even inside the face.. tried both with mouth and smile.

I tried foreach coords and check the one with the high confidence also false positive
here is demo of one of my tires
http://bit.ly/1PGOEYG
what I am missing? thanks

i did below changes to video

after that sunglasses show at wrong position why so have i to make changes at other place also ??

Hi!
We have an web application and there we are currently using Headtrackr.js library for face detection. We want to do face detection with pre-recorded video stream. Headtrackr library provides us such feature to either use webcam for real time face detection or pre-recorded video stream. Now we just want to use js-objectdetect library and trying to integrate it with pre-recorded video stream. Is that possible through this library to either use webcam for real time face detection or with pre-recorded video stream? We will really appreciate your consideration for this issue .

Thanks

Hi, I'm trying to detect a hand open using the hand down in the picture.. Is possible to track the hand in 180 degrees, with fingers pointing to down?

The js-objectdetect can work with rotation limits in degrees? I'm planning to rotate the canvas before find, but it will be so expensive computationally... I really don't now what's the best way to find a hand down nested to the body... I'm trying to track the hand in the following positions:

The images are merely illustrating the intent of each human position. I'm trying to track real hands from people in the web camera.

Hi @mtschirs I love your project.
Would you mind I create a project using your hand gesture code?

I want to make it available to others, as a javascript sdk.

My env:

Browser: Chrome  63.0.3239.132
OS: MacOS Sierra 10.12.6

Hi,

We are testing js-objectdetect as part of an application to measure activity patterns. I have some questions that I would like to ask directly to you. Can you please email me at daniel(dot)zakrisson(a)qbtech(dot)com?

Best regards,
Daniel

I'm trying to make a smiledetector, for http://igsmilebox.blogspot.com/, how do I go about determining if a mouth is smiling? Any help would be much appreciated.