I download a model from the SSD branch of Caffe which is an VGGNet model trained with the COCO dataset. (With 81 classes). I compiled the model using mvNCCompile and it compiled successfully. Then I used this code to run inference:
However, I get the following error:
E: [ 0] dispatcherEventReceive:200 dispatcherEventReceive() Read failed -1
On this line:
(output, _) = output_fifo.read_elem()
@Samuel I was able to reproduce the issue you were seeing with the model you provided and I've previously come across a similar issue with a different network. For the other network, the users were able to resolve their issue by reducing their input size. We are still investigating the root cause of this issue. I'll report back to this thread if I find anything. Thanks.
@Tome_at_Intel Thanks for assisting on this. I really need object detection to work on specific class on Movidius.My initial thought was that it is a bug with the NSDK2 so I downloaded and compiled NSDK1. I recompiled the model and now it does not show any error during real-time detection. However, it shows hundreds of boxes with confidence lower than 2%