Execute deastream as a model that I trained

How do I use my trained model if I use a deepstram to make a real-time inference? I retrain ssd mobileenet v2 following the method of object detection api. A total of 10,000 steps were carried out and model.ckpt was obtained. The Deepstram tutorial does not provide real-time detection of models trained with my data.
model.JPG

Hi gpqls7669,
May I know how did you “retrain ssd mobileenet v2”? Is it TLT? The 1.0.1 TLT does not support ssd with mobilenet_v2 backbone.

I’m sorry for the late reply. I followed the object detection tutorial from tensorflow.