Post Processing for the Object Detection API while converting model from its checkpoint to tflite

Without the help of the post-processing layer, we wanted to convert the raw results to useable bounding boxes ourselves
Your environment

ubuntu 16.04
Tensorflow 1.14.0

Issue

While converting checkpoints to tflite format. As due to micro-controller compatibility, we don't want to use post-processing. Details are as:

post-processing is true: frozen graph adds a
TFLite_Detection_PostProcess custom op node has four outputs:
detection_boxes: a float32 tensor of shape [1, num_boxes, 4] with box
locations
detection_classes: a float32 tensor of shape [1, num_boxes]
with class indices
detection_scores: a float32 tensor of shape [1, num_boxes]
with class scores
num_boxes: a float32 tensor of size 1 containing the number of detected boxes
else:
the graph has two outputs:
'raw_outputs/box_encodings': a float32 tensor of shape [1, num_anchors, 4]
containing the encoded box predictions.
'raw_outputs/class_predictions': a float32 tensor of shape
[1, num_anchors, num_classes] containing the class scores for each anchor
after applying score conversion.

So issue is how to decode the raw_outputs in order to get the processed bounding boxes???