[TRT] attempting to open engine cache file .2.1.GPU.FP16.engine

Autonomous Machines Jetson & Embedded Systems Jetson TX2
#1

I’ve recently re-installed all NVCaffe, DIGITS, Jetson-inference and the environment seems to have changed slightly such that I can no longer run the following without errors:

$ cd jetson-inference/build/aarch64/bin
$ NET=DetectNet-COCO-Dog
$ ./detectnet-camera \
--prototxt=$NET/deploy.prototxt \
--model=$NET/snapshot_iter_38600.caffemodel \

It seems that, although I specify: NET=DetectNet-COCO-Dog, there’s no search for a cache file at this location.

Detectnet-camera works fine with the following, which should have produced a cache somewhere?

$ cd jetson-inference/build/aarch64/bin 
$ ./detectnet-camera coco-dog

Here’s the full error … Please advise!

nvidia@tegra-ubuntu:~$ cd jetson-inference/build/aarch64/bin
nvidia@tegra-ubuntu:~/jetson-inference/build/aarch64/bin$ NET=DetectNet-COCO-Dog
nvidia@tegra-ubuntu:~/jetson-inference/build/aarch64/bin$ ./detectnet-camera \
> --prototxt=$NET/deploy.prototxt \
> --model=$NET/snapshot_iter_38600.caffemodel \
> 
detectnet-camera
  args (3):  0 [./detectnet-camera]  1 [--prototxt=DetectNet-COCO-Dog/deploy.prototxt]  2 [--model=DetectNet-COCO-Dog/snapshot_iter_38600.caffemodel]  

[gstreamer] initialized gstreamer, version 1.8.3.0
[gstreamer] gstCamera attempting to initialize with GST_SOURCE_NVCAMERA
[gstreamer] gstCamera pipeline string:
nvcamerasrc fpsRange="30.0 30.0" ! video/x-raw(memory:NVMM), width=(int)1280, height=(int)720, format=(string)NV12 ! nvvidconv flip-method=0 ! video/x-raw ! appsink name=mysink
[gstreamer] gstCamera successfully initialized with GST_SOURCE_NVCAMERA

detectnet-camera:  successfully initialized video device
    width:  1280
   height:  720
    depth:  12 (bpp)


detectNet -- loading detection network model from:
          -- prototxt     DetectNet-COCO-Dog/deploy.prototxt
          -- model        DetectNet-COCO-Dog/snapshot_iter_38600.caffemodel
          -- input_blob   'data'
          -- output_cvg   'coverage'
          -- output_bbox  'bboxes'
          -- mean_pixel   0.000000
          -- class_labels NULL
          -- threshold    0.500000
          -- batch_size   2

[TRT]  TensorRT version 4.0.2
[TRT]  desired precision specified for GPU: FASTEST
[TRT]  requested fasted precision for device GPU without providing valid calibrator, disabling INT8
[TRT]  native precisions detected for GPU:  FP32, FP16
[TRT]  selecting fastest native precision for GPU:  FP16
[TRT]  attempting to open engine cache file .2.1.GPU.FP16.engine
[TRT]  cache file not found, profiling network model on device GPU
[TRT]  device GPU, loading  
[TRT]  CaffeParser: Could not open file 
[TRT]  CaffeParser: Could not parse model file
[TRT]  device GPU, failed to parse caffe network
device GPU, failed to load 
detectNet -- failed to initialize.
detectnet-camera:   failed to initialize imageNet
nvidia@tegra-ubuntu:~/jetson-inference/build/aarch64/bin$
#2

cd jetson-inference/build/aarch64/bin NET=networks/DetectNet-COCO-Dog
$ ./detectnet-camera
–prototxt=$NET/deploy.prototxt
–model=$NET/snapshot_iter_38600.caffemodel \

#3

Hi,

The log indicates that the application cannot find the model path correctly.
As you already found, you can solve this issue by setting the $NET parameter:
https://github.com/dusty-nv/jetson-inference#loading-custom-models-on-jetson

Thanks.

#4

Hello I am trying to implement jetson-inference tutorial but I after I got to the building your rpo from source i got this error

[TRT]  TensorRT version 3.0
[TRT]  detected model format - caffe  (extension '.caffemodel')
[TRT]  desired precision specified for GPU: FASTEST
[TRT]  requested fasted precision for device GPU without providing valid calibrator, disabling INT8
[TRT]  native precisions detected for GPU:  FP32, FP16
[TRT]  selecting fastest native precision for GPU:  FP16
[TRT]  attempting to open engine cache file .2.1.GPU.FP16.engine
[TRT]  cache file not found, profiling network model on device GPU
[TRT]  device GPU, loading  
could not open file

can you help me

#5

Hi,

Please follow this page to build jetson_inference from source:
https://github.com/dusty-nv/jetson-inference/blob/master/docs/building-repo.md#building-the-repo-from-source

Based on your log, the model is not well-downloaded in your environment.
Have you executed the cmake command successfully?

Thanks.

#6

Hi, AastaLLL

thanks for replying.

yes, I have redownloaded and reinstall the environment but it didn’t work. and the cmake error didn’t show any error.

but when I tried with my own datasets both the image classification and object detection model worked. but it didn’t work with the model that used in the tutorial.

thanks.

#7

Hi aosanshugaa, what command are you running to launch the program from terminal?

#8

hi dusty_nv
when i try the example from the tutorial it doesn’t work

./imagenet-camera googlenet

or

./detectnet-camera  \
--prototxt=$NET/deploy.prototxt \
--model=$NET/snapshot_iter_13110.caffemodel \
--labels=$NET/labels.txt \
--input_blob=data \
--output_blob=softmax

it shows this error

[TRT]  TensorRT version 3.0
[TRT]  detected model format - caffe  (extension '.caffemodel')
[TRT]  desired precision specified for GPU: FASTEST
[TRT]  requested fasted precision for device GPU without providing valid calibrator, disabling INT8
[TRT]  native precisions detected for GPU:  FP32, FP16
[TRT]  selecting fastest native precision for GPU:  FP16
[TRT]  attempting to open engine cache file networks/bvlc_googlenet.caffemodel.2.1.GPU.FP16.engine
[TRT]  cache file not found, profiling network model on device GPU
[TRT]  device GPU, loading  networks/bvlc_googlenet.caffemodel
could not open file Could not parse deploy file
[TRT]  device GPU, failed to parse caffe network
device GPU, failed to load networks/bvlc_googlenet.caffemodel
failed to load networks/bvlc_googlenet.caffemodel
imageNet -- failed to initialize.
imagenet-console:   failed to initialize imageNet

but when i try with my own datasets it works

#9

Hi,
I have exactly the same problem with segnet-console using Pascal-VOC dataset.
I run :

./segnet-console test.jpg output_0428.png \
--prototxt=$NET/deploy.prototxt \
--model=$NET/snapshot_iter_8790.caffemodel \
--labels=$NET/pascal-voc-classes.txt \
--input_blob=data \ 
--output_blob=score_fr

And I get :

[TRT]  TensorRT version 5.0.6
[TRT]  detected model format - caffe  (extension '.caffemodel')
[TRT]  desired precision specified for GPU: FASTEST
[TRT]  requested fasted precision for device GPU without providing valid calibrator, disabling INT8
[TRT]  native precisions detected for GPU:  FP32, FP16
[TRT]  selecting fastest native precision for GPU:  FP16
[TRT]  attempting to open engine cache file .2.1.GPU.FP16.engine
[TRT]  cache file not found, profiling network model on device GPU
[TRT]  device GPU, loading  
[TRT]  CaffeParser: Could not open file 
[TRT]  CaffeParser: Could not parse model file
[TRT]  device GPU, failed to parse caffe network
device GPU, failed to load 
segNet -- failed to initialize.
segnet-console:   failed to initialize segnet

I’m also pretty sure to have properly built jetson-inference, Do you have any idea to solve this ?