Different execution environment in Yolov4 and faster r-cnn

Hello guys, i am running two object detectors: Yolov4 and faster r-cnn… after running yolov4, in output there is written that started parallel pool with 4 workers… after running faster r-cnn there is written running on single gpu… i didnt specify the execution environment in training options, so why one of them uses single gpu and other use something else?

Answer 1

[Vivek Akkala]

The DispatchInBackground parameter within the trainingOptions is accountable for allowing the parallel pool. The Object Detection Using YOLO v4 Deep Learning example has the DispatchInBackground parameter set to true, while Object Detection Using Faster R-CNN Deep Learning example has the parameter set to the default value, which is false. As a result, the parallel pool code is only written in the YOLO v4 example execution.