Tensorflow' pb and pbtxt files don't work with OpenCV after retraining MobileNet SSD V1 COCO

Question

I have followed this tutorial to retrain MobileNet SSD V1 using Tensorflow GPU as described and got 0.5 loss after training using GPU (below more info about config) and

Answer 1

The error was caused by the wrong input .pbtxt file passed into the function readNetFromTensorflow because the .pbtxt has to be geneated by tf_text_graph_ssd.py as describe here:

Run this script to get a text graph of SSD model from TensorFlow Object Detection API. Then pass it with .pb file to cv::dnn::readNetFromTensorflow function.

For other models such as faster r-cnn and mask r-cnn, there are also corresponding scripts.

PS: I just found there is a very good official tutorial here.