Hello Mr.Chandra,
Glad you find it useful and regarding your query, by default during inference, the model itself maps the bbox co-ordinates and provides the output image with bbox in it.
Since you wanted to extract the bbox co-ordinates in a separate file (.txt), I have mentioned the steps in the post itself.
Search for "Interpret outputs from the .txt file : (Optional Read)" section in the post, where I have mentioned codes to extract bbox co-ords into a .txt file for a given image and later denormalize the contents in the .txt file and manipulate the values to proper format to use opencv's patches.rectngle function on it.
Bye,
BALA