CODEWITHSUNDEEP
androidimage-processingcomputer-visiontensorflow-litemulticlass-classification
Sachin
Sachin

Reputation: 1

TensorFlow Lite - Custom model getting replaced by example model

I am learning TensorFlow lite by building a binary Image Classification App for Android. I trained my tflite model by following the official model maker tutorial (https://www.tensorflow.org/lite/tutorials/model_maker_image_classification), where I got my model.tflite and label.txt file saved in the google colab directory. As per the TensorFlow lite tutorial (https://www.tensorflow.org/lite/models/image_classification/overview), I added model.tflite and label.txt (containing the labels of my classes) in assets folder on the reference android app. And changed the label filename in code and TensorFlow Lite file name in code of the reference android app.

However, while running the reference app in Android studio, I am still seeing the default model provided by TensorFlow using the default classes for classification, and not the model I trained and classes of classification I created.

Upvotes: 0

Views: 1306

Answers (1)

Smaa ElNagar
Smaa ElNagar

Reputation: 11

go to download_model.gradle each time the app is built, the gradle download the cefault model replace it with your model link (upload your model to google drive and put the link in the gradle)

Upvotes: 0

Related Questions