I am looking to develop an application that will allow the user to launch the camera and use it to scan what the camera sees for a certain logo or word.
I have had some experience using the Zebra barcode scanning library so I am able to launch the camera and scan for a barcode fine.
The issue comes when I try to scan for a logo. Is it at all possible to scan for the logo in real time with the camera or would I have to have the user take picture and then scan the newly take picture and compare it with the logo that is being searched for?
Scanning for a logo is an entirely different problem. The Zebra Library has been designed specifically to recognized barcodes and qr codes anything outside of that scope and it won't be able to interpret anything. You can scan for logos in real time or on the cloud. Amazon does this with their shopping application so that you can research what you find at the supermarket or at a friends house.
There are companies that can offer you Image Recognition as a Service:
https://www.clarifai.com
or you can look into your own open source implementation
opencv.org/
There's examples for OpenCV for Android but you're going to need a bit of background on Machine Learning and Computer Vision before you get started.
Related
We have a few Android apps that utilize com.google.android.gms:play-services-vision OCR library and it seem to work fine in many countries, but for some odd reason is not activated in Sudan. The app directs the video stream from camera to Vision, receives recognized strings, and displays them in an overlay on top of the video stream. Sudan is the only country where Vision OCR fails out of many where the app works just fine.
Any idea what could be the reason?
Example app: https://play.google.com/store/apps/details?id=com.arl.shipping.gateexecutor, install it, fill a form to get a pairing code and click an icon with shipping container. This should invoke an OCR that would look for a valid shipping container number in the camera view.
I am working on app that detect eye blink of the user. I have been searching the web for 2 days but still don't have clear vision about how this can be done.
As far as i have knew is that the system supports face detection which is detecting if there is a face in the picture and locating it.
But this works only with images and detect only faces which is not what i need. I need to open an camera activity and directly detect the face of the user and locate his eyes and other facial parts and wait till he blinks, like when you long click on the screen on snap chat.
I have seen a lot about open-cv but still not sure what it is or how to use it or if it seize my goals.
Note: snap chat has no API released for the technology used, and even it doesn't let anyone to talk to the engineers behind this technology.
I know that openCV has the ability to allow image processing on the device's camera feed (as opposed to only being able to process still images).
Here is an introductory tutorial on eye detection using openCV:
http://romanhosek.cz/android-eye-detection-and-tracking-with-opencv/
If you can't find eye-blink detection tutorials in a google search, I think you'll have to create the code for eye-blink detection on your own, but I think openCV will be a helpful tool in doing so. There are lots of beginner openCV tutorials to help you get started.
I was wondering if there are any apps/(Xposed) modules which can force an image to be seen by the camera.
Basically, I'd like a scanner to scan a self-made barcode rather than needing two phones; one to generate and show the barcode and the other to scan it from a specific app.
Anything that will show an image for the scanner in an app will do.
I am developing an app which takes photos with the use of the voice. I have read the google sdk voice recognizer, but I haven't quite understood it. So here I am asking, what specifically can I add to my app with the use of the camera that will tell my app to listen for the word 'snap' in order to take a photo. I haven't found a real tutorial of it nor do I understand how to really code it by just looking in the android sdk. I want to be able to open the camera, and then you click the listening button, and it waits for the user to say 'snap' then takes a photo, review it, then return back to the camera viewer.
Am currently trying out zxing to develop a MultiFormat Barcode reading supporting android platform, To get Started have downladed zxing 1.7 and builded.
It works fine with EAN format codes and QR Codes, But It fails to scan a Barcode when it comes to real scanning of products which has small UPC barcodes.
Is this a problem with device?
Am Testing it on my Samsung GT-S5570, Android Version 2.2.1, With Autofocus enabled Camera 3.1 MP.
Kindly advice.
Thanks
Sam
If you can scan EAN barcodes, but not small UPC barcodes, it is probably because the camera can't get enough detail to distinguish the barcode from the background. Here are some things you could try:
Hold the camera closer to, or further away from, the barcode. Too close, and the auto focus mechanism is unable to compensate. Too far away, and the barcode becomes too small for detection.
Use more light. ZXing can activate the flash-light on some phones, but not all. If the lightning conditions are too poor, the contrast between the background and the barcode becomes too small.
Try with a different phone, with different/better camera.