Community
cancel
Showing results for 
Search instead for 
Did you mean: 
Highlighted
Community Manager
194 Views

[caffe] Simple example does not compile ( bias_term: false)

I've got very simple example with only 4 nodes (depth-wise and point-wise convs).

 

But I'm getting compile error.

 

mvNCCompile -s 12 op.prototxt -in MobilenetV1_Conv2d_0_Conv2D mvNCCompile v02.00, Copyright @ Movidius Ltd 2016 ****** WARNING: using empty weights ****** Fusing depthconv and conv in MobilenetV1_Conv2d_1_depthwise_depthwise and MobilenetV1_Conv2d_1_pointwise_Conv2D Traceback (most recent call last): File "/usr/local/bin/mvNCCompile", line 118, in <module> create_graph(args.network, args.inputnode, args.outputnode, args.outfile, args.nshaves, args.inputsize, args.weights) File "/usr/local/bin/mvNCCompile", line 101, in create_graph net = parse_caffe(args, myriad_config) File "/usr/local/bin/ncsdk/Controllers/CaffeParser.py", line 1394, in parse_caffe network.attach(node) File "/usr/local/bin/ncsdk/Models/Network.py", line 81, in attach stage.attach_several(appropriate_nodes) File "/usr/local/bin/ncsdk/Models/NetworkStage.py", line 695, in attach_several parents.attach(self) File "/usr/local/bin/ncsdk/Models/NetworkStage.py", line 418, in attach taps[c,c*multiplier+i,y,x] = self.taps[y,x,c,i] IndexError: index 3 is out of bounds for axis 2 with size 3

 

op.prototxt:

 

layer { name: "image" type: "Input" top: "image" input_param { shape { dim: 1 dim: 3 dim: 432 dim: 368 } } } layer { name: "MobilenetV1_Conv2d_0_Conv2D" type: "Convolution" bottom: "image" top: "MobilenetV1_Conv2d_0_Conv2D" convolution_param { num_output: 24 bias_term: false kernel_size: 3 group: 1 stride: 2 pad_h: 1 pad_w: 1 } } layer { name: "MobilenetV1_Conv2d_1_depthwise_depthwise" type: "Convolution" bottom: "MobilenetV1_Conv2d_0_Conv2D" top: "MobilenetV1_Conv2d_1_depthwise_depthwise" convolution_param { num_output: 24 bias_term: false kernel_size: 3 group: 24 stride: 1 pad_h: 1 pad_w: 1 } } layer { name: "MobilenetV1_Conv2d_1_pointwise_Conv2D" type: "Convolution" bottom: "MobilenetV1_Conv2d_1_depthwise_depthwise" top: "MobilenetV1_Conv2d_1_pointwise_Conv2D" convolution_param { num_output: 48 bias_term: false kernel_size: 1 group: 1 stride: 1 pad_h: 0 pad_w: 0 } }

 

I've noticed that if I remove bias_term: false from MobilenetV1_Conv2d_1_depthwise_depthwise it compiles without errors.

 

What's wrong with bias_term: false ?
Tags (1)
0 Kudos
1 Reply
Highlighted
Community Manager
8 Views

@jokilokis We have support for Tiny Yolo in the NCAPPZOO (https://github.com/movidius/ncappzoo/tree/master/caffe/TinyYolo) and it uses the parameters (bias_term=true & bias_term=false) without any issues.

0 Kudos