Hello, I am engaged on OpenVINO coaching extensions to coach object detection mannequin with customized dataset.

I efficiently educated mannequin however I discovered the output layer solely contains packing containers and labels.

I want confidence rating to implement my software.

Earlier than making an attempt to make use of customized mannequin, I used ssdlite_mobilenet_v2. And this mannequin’s output is  DetectionOutput (every detection has the format: [image_id, label, conf, x_min, y_min, x_max, y_max]). 

I ponder why the default construction for object detection(in openvino coaching extensions) mannequin would not have a confidence rating.

I’ve looking associated paperwork and code however have not discovered reply..

How can I modify this construction to incorporate confidence rating?

Thanks.