I'm trying to develop a mobile app for traffic sign recognition and i want it to be in real-time. I'm trying only to detect circles signs and to find out what sign is in order to notify the driver. I want to know what is the best method I should use. For now I've tried using java and opencv to find the circles in an image (using HoughCircles) but is not quite what I've expected - a lot of signs aren't identified. Then I tried to use opencv for training it to learn the signs - to obtain an xml trained classifier, but it takes too long and to be functionally I need a really large amount of data. I don't know what to do ... Thank you in advance.
I found this work with some research: https://www.academia.edu/4950526/Traffic_Sign_Recognition_system_on_Android_devices
Related
I want to integrate Face Recognition in Android.
1) First time user will sign up by scanning face (Will stored on Server/local)
2) Next time when a user tries, it should detect the user from the stored image.
I tried exploring Face Recognition of AWS but ended up with confusion in setting up.
there are multiple ways you can do this.
there is a simple library for this purpose in android called FaceDetector. you can use that for face detection only.
there are also more advanced libraries like OpenCV which is available for many languages including java and android. but I recommend if you are not familiar with android NDK or computer vision try sticking with the FaceDetector library.
note that if you start with OpenCV. it is going to be hard for you at first because you are entering a field called computer vision which is a branch of machine learning. and immediately you will feel it that you're walking in a completely new territory
I am a long time lurker on Stack Overflow where generally the solutions I sought already existed on here. However, I am stumped trying to find information on tackling this problem.
Basically, I am looking at making an Android app - first time making such an app - and part of the functionality is to monitor what else the phone user is doing and then maybe create a log of what the person has done. For example, spent 27% of the time when the phone screen is on using Facebook.
As for my actual Android Studio experience, all I have created so far is a simple app that takes in text from an input field and alerts it back out to you. I am studying a video series to learn how to use and develop with Android Studio.
I am enthusiastic about the challenge, so if anyone has some good resources or information to help me achieve this I would be really grateful.
Yes it is possible to track applications usage, using google API.
Here is google resource link
What may interest you the most is the part concerning UsageStats
You can use this since api 5.0. You can then for example use method getTotalTimeInForeground() to get the total time this package spent in the foreground, measured in milliseconds, which seems to correspond to your needs.
Therefore, using your Facebook example. After you get the package name (com.facebook.katana) , you could get the application (Facebook) uptime on a daily/weekly/monthly base.
Also, you could use getLastTimeUsed() and make your application logs those timestamp then do some maths.
I want to do Face recoginition (Not Face Detection) in my Android and iOS app. I have studied a lot on Web and found following possible solutions:
1.) openCV: I don't want to go into writing my own API using this. Also, I don't have prior experience in JNI for Android.
2.) Betaface API So far this is good.
3.) Sky Biometrics is also Good.
Now, I am searching for the solution from 3-5 days and came to know that I can use above API (so far I have decided to purchase license for Sky Biometrics). And this API will provide me a list of Features for the faces it recognised.
But, Now I am confused That how to use these features and save in my local data base to recognise faces from the pictures.So My queries are following
1.) How to convert Face features to Actual working Face recognition API means What is the actual algo or solution I can use to merge diffrent face features of a same person to identify him correctly.
2.) Uploading images and then creating database for Face-features set is a very time taking process. Do any one knows any Android/iOS Face Recoginition SDK to do this so that I can do this accurately and timely with no or less time taking process.
3.) Both solution-2 and 3 can be used with Images. Is there any other solution is available which can do the same with less efforts but with more accuracy.
OpenBR may be also interesting for you: http://openbiometrics.org/
Finally I am using Rekognition API. And this is good enough to serve my purpose.
I have to develop an android application able to acquire data and chart them in real time.
Now, on Android Platform there are not built-in graphical libraries, however It is open source, so there are many libraries you can download and install.
Many of these are not able to perform real time chart, and the most part lacks of tutorial and code samples, as a consequence I don't know which I could choose.
I have just tested Android Plot but the graphic thread clashed with the thread of acquisition and the main thread, causing a deadlock and I didn't know why.
I found on a web forum that this library doesn't fit very well with real time chart, so I decide to give up it.
I would be grateful if you advice me which library is more suitable for my problem.
I have never tried this although a technically possible solution would be to use AJAX-y requests in combination with the Google Charts Query Wrapper?
Could be awesome but this means your app would only be able to chart when connected to the internet. Where is the data being stored? Online or in a local SQLite database? The query wrapper chart would be less appropriate if you were just reading your data out of a local database but the approach (AJAX & Google Charts) would be sweet.
I am trying to build an android application that allows user to login through signing on a canvas, I can't find any algorithm that will help me implement this functionality in reliable manner , I read alot of papers and methods about signature recognition such as fourier transform method but all the papers involve very complex mathematical operations which makes it very hard to implement using the android sdk.
so is there any open source algorithm that would compare two signature bitmaps(drawables) to obtain descent results?
if not how should I go about this issue?
Comparing drawables may be not sufficient, as signatures may differ in size and position. You can attempt recognition based on invariant moments ( I use this technology for OCR,
but on printed characters, see demos in http://sourceforge.net/projects/javaocr/ project )
You may get better results analysing signature process in dynamics - path, speed, energy etc.
I don't know anything about it but I feel there is sample app installed in the emulator called GestureBuilder which can help you. You can refer Gesture for this.