Community
cancel
Showing results for 
Search instead for 
Did you mean: 
idata
Community Manager
715 Views

Yolov3

I try to convert yolov3-tiny to caffe. The prototxt like https://github.com/Amalle/Yolov3_caffe/blob/master/models/caffe/yolov3-tiny.prototxt

 

And I convert the graph using the following:

 

$ mvNCCompile models/caffe/yolov3-tiny.prototxt -w models/caffe/yolov3-tiny.caffemodel -s 12 -ec mvNCCompile v02.00, Copyright @ Intel Corporation 2017 /usr/local/bin/ncsdk/Controllers/FileIO.py:65: UserWarning: You are using a large type. Consider reducing your data sizes for best performance Blob generated

 

But the predicted result size was wrong.

 

Then I try mvNCProfile:

 

$ mvNCProfile models/caffe/yolov3-tiny.prototxt -w models/caffe/yolov3-tiny.caffemodel -ec mvNCProfile v02.00, Copyright @ Intel Corporation 2017 /usr/local/bin/ncsdk/Controllers/FileIO.py:65: UserWarning: You are using a large type. Consider reducing your data sizes for best performance Blob generated USB: Transferring Data... /usr/local/lib/python3.5/dist-packages/mvnc/mvncapi.py:420: DeprecationWarning: The binary mode of fromstring is deprecated, as it behaves surprisingly on unicode inputs. Use frombuffer instead Time to Execute : 266.8 ms USB: Myriad Execution Finished Time to Execute : 248.16 ms USB: Myriad Execution Finished Traceback (most recent call last): File "/usr/local/bin/mvNCProfile", line 156, in <module> profile_net(args.network, args.inputnode, args.outputnode, args.nshaves, args.inputsize, args.weights, args.device_no, args.explicit_concat, args.ma2480, args.scheduler, args) File "/usr/local/bin/mvNCProfile", line 142, in profile_net timings, myriad_output = run_myriad(graph_file, args) File "/usr/local/bin/ncsdk/Controllers/MiscIO.py", line 248, in run_myriad myriad_output = myriad_output.reshape(sz[1], sz[2], sz[3]) ValueError: cannot reshape array of size 41308 into shape (1,1,215475)

 

@Tome_at_Intel
Tags (3)
0 Kudos
4 Replies
idata
Community Manager
182 Views

@maqiao Looks like this Tiny Yolo v3 uses a concat as the last layer. There is a known issue where the NCSDK has issues with models having concat as the last layer. Some users have found work arounds by using dummy reshape layers https://ncsforum.movidius.com/discussion/comment/2831/#Comment_2831. Try this and let me know if this works for you.

idata
Community Manager
182 Views

@Tome_at_Intel Thanks

 

I have try that. It can't works for me.

 

The last three layers as following:

 

layer { name: "conv13_perm" type: "Permute" bottom: "conv13" top: "conv13_perm" permute_param { order: 0 order: 2 order: 3 order: 1 } } layer { name: "conv13_flat" type: "Flatten" bottom: "conv13_perm" top: "conv13_flat" flatten_param { axis: 1 } } layer { name: "output" type: "Concat" bottom: "conv13_flat" bottom: "conv10_flat" top: "output" concat_param { axis: 1 } }

 

Try this:

 

`

 

$ mvNCProfile models/caffe/yolov3-tiny.prototxt -w models/caffe/yolov3-tiny.caffemodel -s 12 -on conv13_perm -ec

 

mvNCProfile v02.00, Copyright @ Intel Corporation 2017

 

/usr/local/bin/ncsdk/Controllers/FileIO.py:65: UserWarning: You are using a large type. Consider reducing your data sizes for best performance

 

Blob generated

 

USB: Transferring Data…

 

Time to Execute : 216.87 ms

 

USB: Myriad Execution Finished

 

Time to Execute : 197.8 ms

 

USB: Myriad Execution Finished

 

USB: Myriad Connection Closing.

 

USB: Myriad Connection Closed.

 

Network Summary

 

Detailed Per Layer Profile

 

Bandwidth time

 

Name MFLOPs (MB/s) (ms)

 

0 data 0.0176147.1 0.006

 

1 scale1 149.5 1175.5 7.582

 

2 pool1 2.8 792.4 6.666

 

3 scale2 398.7 859.5 13.839

 

4 pool2 1.4 926.3 2.851

 

5 scale3 398.7 648.0 9.226

 

6 pool3 0.7 957.8 1.379

 

7 scale4 398.7 403.3 7.717

 

8 pool4 0.3 947.6 0.697

 

9 scale5 398.7 203.8 10.051

 

10 pool5 0.2 924.2 0.357

 

11 scale6 398.7 323.6 9.250

 

12 pool6 0.4 411.4 0.401

 

13 crop6 0.0 2180.1 0.088

 

14 scale7 1594.9 314.8 33.309

 

15 scale8 88.6 331.9 2.501

 

16 scale9 398.7 323.5 9.252

 

17 conv10 44.1 311.0 1.332

 

18 scale11 11.1 441.6 0.328

 

19 scale11_s0 0.0 33.8 0.010

 

20 upsample_p0 0.0 156.4 0.264

 

21 scale11_s1 0.0 38.7 0.009

 

……

 

273 scale11_s127 0.0 38.2 0.009

 

274 upsample_p127 0.0 158.2 0.261

 

275 pool_deconv 0.1 566.3 0.292

 

276 route 0.0 1950.4 0.085

 

277 route_scale5 0.0 2094.7 0.158

 

278 scale12 1196.2 210.3 29.215

 

279 conv13 88.3 197.3 2.305

 

280 conv10_perm 0.0 310.1 0.265

 

281 conv10_flat 0.0 179.6 0.458

 

282 conv13_perm 0.0 182.7 1.800

 

Total inference time 186.13

 

Generating Profile Report 'output_report.html'…

 

`

 

But, the following was wrong:

 

`

 

$ mvNCProfile models/caffe/yolov3-tiny.prototxt -w models/caffe/yolov3-tiny.caffemodel -s 12 -on conv13_flat -ec

 

mvNCProfile v02.00, Copyright @ Intel Corporation 2017

 

/usr/local/bin/ncsdk/Controllers/FileIO.py:65: UserWarning: You are using a large type. Consider reducing your data sizes for best performance

 

Blob generated

 

USB: Transferring Data…

 

Segmentation fault (core dumped)

 

`
idata
Community Manager
182 Views

I trying to use yolov3-tiny model for Movidius but i checked release note for ncsdk v2.08.01 yolov3 is not being enlisted in the list of supported models.

 

Does it means i cannot use yolo v3 model with movidius stick with new ncsdk?

 

I tested every possible way in which i can use yolov3 tiny model with movidius stick i was able to get the prototxt file and caffemodel as well.But it don't work.

 

Is there any possible way i can use the yolov3-tiny model with movidius stick?
idata
Community Manager
182 Views

Reply