frame

Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!

Sign In

Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!

Tiny YOLO on NCS

I've been trying to find a way to run an object detection network on the NCS. After a while trying and converting different models I finally found a one that seems to work; Tiny YOLO. I haven't yet found a way to convert the weights from darknet to caffe to NCS, but with empty weights the performance looks promising. Under 150ms per image!

python3 mvNCProfile.pyc yolo_tiny_deploy.prototxt -s 12
mvNCProfile v02.00, Copyright @ Movidius Ltd 2016

****** WARNING: using empty weights ******
USB: Transferring Data...
Time to Execute : 140.52 ms
USB: Myriad Execution Finished
Time to Execute : 121.39 ms
USB: Myriad Execution Finished
USB: Myriad Connection Closing.
USB: Myriad Connection Closed.
Network Summary

Detailed Per Layer Profile

Layer Name MFLOPs Bandwidth MB/s time(ms)

0 scale1 173.408 1173.26 8.81
1 pool1 3.211 811.31 7.55
2 scale2 462.422 964.80 14.30
3 pool2 1.606 956.24 3.20
4 scale3 462.422 698.33 9.92
5 pool3 0.803 1008.54 1.52
6 scale4 462.422 428.05 8.38
7 pool4 0.401 1001.39 0.76
8 scale5 462.422 205.36 11.13
9 pool5 0.201 980.96 0.39
10 scale6 462.422 308.66 10.08
11 pool6 0.100 944.83 0.20
12 scale7 462.422 809.99 11.64
13 scale8 231.211 596.90 8.98

14 fc9 0.025 2146.64 16.40

Total inference time 113.27

Does anyone have experience converting the weights? Or knowledge how to draw the bounding boxes based on the YOLO output layer? It appears to output 1470 values.

Comments

Sign In or Register to comment.