TensorRT Sample "sampleSSD" error

AI & Data Science Deep Learning (Training & Inference) TensorRT
1

ubuntu16.04
tensorRT :TensorRT-5.0.2.6.Ubuntu-16.04.4.x86_64-gnu.cuda-9.0.cudnn7.3
caffemodel:VGG_VOC0712_SSD_300x300_iter_120000.caffemodel
prototxt:deploy.prototxt
(models_VGGNet_VOC0712_SSD_300x300.tar.gz from GitHub - weiliu89/caffe at ssd → models_VGGNet_VOC0712_SSD_300x300.tar.gz - Google Drive according to TensorRT-5.0.2.6/samples/sampleSSD/README.md)

I run sampleSSD sample in TensorRT-5.0.2.6/bin/sampleSSD,the error is as follows:

user@2GPU:~/code/tensorRT/TensorRT-5.0.2.6/bin$ ./sample_ssd
Begin parsing model…
FP32 mode running…
Warning: Flatten layer ignored. TensorRT implicitly flattens input to FullyConnected layers, but in other circumstances this will result in undefined behavior.
Warning: Flatten layer ignored. TensorRT implicitly flattens input to FullyConnected layers, but in other circumstances this will result in undefined behavior.
Warning: Flatten layer ignored. TensorRT implicitly flattens input to FullyConnected layers, but in other circumstances this will result in undefined behavior.
Warning: Flatten layer ignored. TensorRT implicitly flattens input to FullyConnected layers, but in other circumstances this will result in undefined behavior.
Warning: Flatten layer ignored. TensorRT implicitly flattens input to FullyConnected layers, but in other circumstances this will result in undefined behavior.
Warning: Flatten layer ignored. TensorRT implicitly flattens input to FullyConnected layers, but in other circumstances this will result in undefined behavior.
Warning: Flatten layer ignored. TensorRT implicitly flattens input to FullyConnected layers, but in other circumstances this will result in undefined behavior.
Warning: Flatten layer ignored. TensorRT implicitly flattens input to FullyConnected layers, but in other circumstances this will result in undefined behavior.
Warning: Flatten layer ignored. TensorRT implicitly flattens input to FullyConnected layers, but in other circumstances this will result in undefined behavior.
Warning: Flatten layer ignored. TensorRT implicitly flattens input to FullyConnected layers, but in other circumstances this will result in undefined behavior.
Warning: Flatten layer ignored. TensorRT implicitly flattens input to FullyConnected layers, but in other circumstances this will result in undefined behavior.
Warning: Flatten layer ignored. TensorRT implicitly flattens input to FullyConnected layers, but in other circumstances this will result in undefined behavior.
ERROR: mbox_loc: all concat input tensors must have the same dimensions except on the concatenation axis
ERROR: mbox_conf: all concat input tensors must have the same dimensions except on the concatenation axis
Caffe Parser: Invalid axis in softmax layer - Cannot perform softmax along batch size dimension and expects NCHW input. Negative axis is not supported in TensorRT, please use positive axis indexing
error parsing layer type Softmax index 98
End parsing model…
Segmentation fault (core dumped)

I think I need to edit the deploy.prototxt file. The tutorial I found is as follows https://docs.nvidia.com/deeplearning/sdk/tensorrt-developer-guide/index.html#samples.
But I don’t know how to change it.

2

Try searching in this thread: Mobilenetssd · Issue #5 · chenzhi1992/TensorRT-SSD · GitHub
It has some discussions on how to modify .prototxt near the end.
Hope it helps

3

Hi,

You have to follow the instructions in the associated README.md with the sampleSSD code for changing deploy.prototxt and then renaming is to ssd.prototxt. The instructions can also be found here: [url]https://docs.nvidia.com/deeplearning/sdk/tensorrt-release-notes/tensorrt-5.html#rel_5-0-2[/url]

I’ve gone through the instructions to produce an example ssd.prototxt file that worked with the sampleSSD code in /workspace/tensorrt/samples/sampleSSD in the TensorRT 5.1 (19.03-pyX) NGC Container: [url]https://ngc.nvidia.com/catalog/containers/nvidia:tensorrt/[/url]

Example ssd.prototxt file after following README.md instructions attached. (You will have to remove the extra .txt extension at the end of the filename.)

Thanks,
NVIDIA Enterprise Support
ssd.prototxt.txt (24.8 KB)

4

Are these changes documented specific to the sample SSD model? How about any other Caffe model? Is there a generic documentation applicable for any model?

5

Hi, The link to download the ssd.prototxt is no longer available. Is it possible to update ?

6

Need example ssd.prototxt…Its not clear to me on what we need to change in the deploy.txt