Intel® Distribution of OpenVINO™ Toolkit
Community support and discussions about the Intel® Distribution of OpenVINO™ toolkit, OpenCV, and all things computer vision-related on Intel® platforms.

FRUSTRUM POINT NET IN NCS

idata
Community Manager
425 Views

Hello,

 

I am facing several issues with my architecture of point net. Which I wish to put on the NCS. Please have a look at the architecture.

 

The get model function defines the model

 

https://github.com/charlesq34/frustum-pointnets/blob/master/sunrgbd/sunrgbd_detection/frustum_pointn...

 

and the tf_util files

 

https://github.com/charlesq34/frustum-pointnets/blob/master/models/tf_util.py

 

issues are with max pool function and conv2d . Moreover I am not sure if ncs supports concat , slice etc. how to proceed with it.

 

or also may consider having a look at the following problem.

 

input was a verctor of shape 1,2048,6,1 and kernel was of size 1,6,1,64

 

print(output_size) ---> output [Dimension(1), 2048, 6, Dimension(64)]

 

print(node.outputs[0]) ----> output Tensor("conv1/Conv2D:0", shape=(1, 2048, 1, 64), dtype=float32)

 

node.outputs[0].set_shape(output_size) ----> in this line of code inside /usr/local/bin/ncsdk/Controllers/TensorFlowParser.py line 370

 

raise ValueError("Shapes %s and %s are not compatible" % (self, other))

 

ValueError: Shapes (1, 2048, 1, 64) and (1, 2048, 6, 64) are not compatible

 

and also in maxpool

 

where input shape :

 

(1, 2048, 1, 1024)

 

and please note I had a kernel size of 2048,1 for max pooling

 

File "/usr/local/bin/mvNCCompile", line 118, in

 

create_graph(args.network, args.inputnode, args.outputnode, args.outfile, args.nshaves, args.inputsize, args.weights)

 

File "/usr/local/bin/mvNCCompile", line 104, in create_graph

 

net = parse_tensor(args, myriad_config)

 

File "/usr/local/bin/ncsdk/Controllers/TensorFlowParser.py", line 700, in parse_tensor

 

args=arguments)

 

File "/usr/local/bin/ncsdk/Models/NetworkStage.py", line 279, in init

 

self.outputDimX))).astype(

 

ValueError: negative dimensions are not allowe
0 Kudos
2 Replies
idata
Community Manager
168 Views

Pleas help as soon as possible

idata
Community Manager
168 Views

Hi

 

tyr to put padding = SAME:

 

net = tf_util.conv2d(input_image, 64, [1,3], padding='SAME', stride=[1,1],
Reply