Local image searching bar (Classify or read the contents of road signs)

Well... We are going to ugrade our software for a competition by next week. We want to make a local searching bar by image and sound but we don't know where to start. Anyone can help us?

What exactly is a "local searching bar by image"? Could you elaborate on what exactly you are looking to do? Do you have screenshots of your app, or a sketch of a prototype?

Well, i mean we want to add a function for our app. For example, when you are on the road, you see a sign and don't know what does it mean. So you open the app, scan the image by camera and the app will show information about that sign.

Hello Jane

You seem to want to use an artificial intelligence to either 'classify' the image as to type or content or to electronically read information about the sign. What you want to do is difficult, even with a professional compiler.

To scan the roadside sign and to show information about that sign, the sign either needs to contain
a QR code (so the sign can be scanned for information electronically ) or the image has to be able to be compared to a 'library' of known images (use an image classifier). It might be possible to use OCR on the image you take of the sign (optical character recognition) to read the words on the sign. Here is a link FAQ Section: Optical Character Recognition (OCR) that might help

An App Inventor extension exists (Personal Image Classifier) that 'works' (kind of, but not really, really well). You can try it. Read about the PIC in these community discussions: https://community.appinventor.mit.edu/search?q=personal%20i

To scan a QR code imbedded on the sign use a Bar Scanner BarcodeScanner . The sign must have a QR code on it somewhere for this technique to work.

Good luck with your Project. If you try anything, post what you tried and someone might be able to provide specific advice.

Edit: Manual of Traffic Road Signs in the US or https://mutcd.fhwa.dot.gov/services/publications/fhwaop02084/index.htm might be helpful to provide 'training' images for image classification

I've tried PIC but it didn't really work. Do you have any tutorials or any example?

Me too. "'works' (kind of, but not really, really well). You can try it." I see it as a curiosity and not really practical yet. Others might have a different experience depending on what they use it for.

Have you done this tutorial https://appinventor.mit.edu/explore/resources/ai/personal-image-classifier which provides advice demonstrating how to use PIC ?

There are several videos https://www.google.com/search?q=app+inventor+personal+image+classifier&rlz=1C1CHBF_enUS887US887&oq=app+inventor+personal+image+classifier&aqs=chrome.0.69i59j69i60l3.10182j0j4&sourceid=chrome&ie=UTF-8 that might help you improve performance.

Results are supposedly tied to the capabilities of your Android device (see the link).

Artificial Intelligence technology is still in its infancy w.r.t. App Inventor.

If you explain what you tried, but didn't really work; someone might be able to give you specific advice. Alternatives might be listed in this google search https://www.google.com/search?q=online+image+classification&rlz=1C1CHBF_enUS887US887&oq=online+image+classification&aqs=chrome..69i57.7551j0j7&sourceid=chrome&ie=UTF-8

I've tried and got the result but I still want to erase these letters which i highlighted in the picture. Do you know how to do that?

You could try Taifun's advice How to parse a result

Or treat that text as a List and segregate the first item (the output appears to be a json list)