you can find how to run faster_rcnn_resnet50_coco in the documentation at the root of the software: computer_vision_sdk_2018.2.299/deployment_tools/documentation/TensorFlowObjectDetectionFasterRCNN.html
As a precision, this model faster_rcnn_resnet50_coco is made for object detection, therefore you will need to run it with objection detection samples such as object_detection_sample_ssd after conversion with the MO.
The object detection is working by running object_detection_sample_ssd, thanks.
But how to do image classification?
Image Classification Sample does not work for it.
image classification is simpler than object detection for image classification you need simpler dnn like alexnet lenet resnet squeezenet ...
In the demo you have sample of image classification with squeezenet.
I am using the Openvino_2020.2.120 for tensoflow, Faster RCNN. What I did was to take the code under the following path /opt/intel/openvino_2020.2.120/deployment_tools/open_model_zoo/demos/mask_rcnn_demo/ and made some small changes to make it work like Faster RCNN and ran this updated inference code using the .xml and .bin files made for Faster RCNN architecture, this worked for me.