Strange problem - most of returned lines from getComponents from recognized texts prefixed with a character

[zbyszek-matuszewski]

About a month ago I had a problem with Google Android Vision OCR when text detector returned items which after getting lines by getComponents() method of TextBlock had a "(' character added at the beginning. It wasn't the case for all the lines though.

Now I have very similar problem but now it is "8". It happens to about 1/3 - 1/2 of recognized lines.
How to approach such problem?

I get texts like that in receiveDetections method:

SparseArray<TextBlock> items = detections.getDetectedItems();
List<Text> all_lines = new ArrayList<>();

for (int i = 0; i < items.size(); ++i) {
    TextBlock item = items.valueAt(i);
    List<? extends Text> texts = item.getComponents();
    for (Text t: texts) all_lines.add(t);
}

This is example how it looks

It is on Samsung Galaxy J5. On other phones it is ok though. It was also ok few hours ago.
Recompiling the app from clean build, reinstaling it, turning off and on again didn't work.

[liuyl]

We didn't change the OCR code after release, so it seems a device specific issue.
I'll try to find a Galaxy J5 to test this, thanks for report.

[huangxuhui]

After you get the all_lines from
for (Text t: texts) all_lines.add(t);
How you add to the mGraphicOverlay?

[zbyszek-matuszewski]

I'm currently not able to find that version of code that produced the result with mGraphicOverlay. But mGraphicOverlay was not important as the same was happening when I logged that value to Log.d(...) and the same results was a problem for our processing in app (which didn't show those text values by mGraphicOverlay but did post-ocr text processing). We've taken String values from t.getValue() where t is an instance of Text class.