Tensorflow Task Library not detecting objects in android app

General Discussion
, ,
#1

I have trained a custom object detection model and exported it to tflite format.
but when I try to use it in my android application it does not detect any objects.
I think the problem is in the below given code snippet.

CustomObjectDetectorOptions options = new CustomObjectDetectorOptions.Builder(localModel)
                .setDetectorMode(ObjectDetectorOptions.STREAM_MODE)
                .setClassificationConfidenceThreshold(0.1f)
                .setMaxPerObjectLabelCount(5)
                .enableClassification()
                .build();

because when I remove the line .enableClassification() from the code it predicts but don’t classify objects.

#3

Hi @Abdullah_Ilyas,

Hope you would have resolved the issue by now. If not, here are a few additional guidelines. enableClassification() should be included in your CustomObjectDetectorOptions to enable the classifications in the live stream/static objects. The issue might be with setClassificationConfidenceThreshold(0.1f) . Already you set the minimum confidence level 0.1. But experiment with the confidence interval between 0.4 - 0.9 for predictions and classifications.

Thank You