difference between `predict` and `adaptive_predict` in text_detection.py

[Horace89]

In the Docker image, we can see that the web server invokes detection_model.predict(bgr_image), and the adaptive_predict function is not used anywhere.
Could you tell what this function is for?