Jetpack 3.2 Tegra_multimedia_api backend sample won't run


I was unable to run the tegra multimedia api backend sample as desribed at the end of this thread: Tegra Multimedia API samples Readme seems to have changed from Jetpack 3.0 to 3.1 - Jetson TX2 - NVIDIA Developer Forums

nvidia@tegra-ubuntu:~/tegra_multimedia_api/samples/backend$ ./backend 1 ~/nvidia/tegra_multimedia_api/data/video/sample_outdoor_car_1080p_10fps.h264 H264 \
> --trt-deployfile ~/nvidia/tegra_multimedia_api/data/model/GoogleNet-modified.prototxt \
> --trt-modelfile ~/nvidia/tegra_multimedia_api/data/model/GoogleNet-modified-online_iter_30000.caffemodel \
> --trt-forcefp32 0 \
> --trt-proc-interval 1 -fps 10
parse net failed, exit!

I changed the odd remaining gie-proc-interval to tie-proc-interval, but still get parse net failed, exit.

Can anyone reproduce this?

Hi sherm_jonathan,
Please check the path to h264, deployfile and modelfile.

The command below is running well:

nvidia@tegra-ubuntu:~/tegra_multimedia_api/samples/backend$ ./backend 1 ../../data/Video/sample_outdoor_car_1080p_10fps.h264 H264 --trt-deployfile ../../data/Model/GoogleNet_one_class/GoogleNet_modified_oneClass_halfHD.prototxt --trt-modelfile ../../data/Model/GoogleNet_one_class/GoogleNet_modified_oneClass_halfHD.caffemodel --trt-forcefp32 0 --trt-proc-interval 1 -fps 10

Indeed it is - thank you!

[b]EDIT: I just needed to have more patience! After a couple of minutes the video came up and it started to work …



I am running Jetpack 3.3 and the above command just hangs with the messages:

Net has batch_size, channel, net_height, net_width:1 3 540 960
forced_fp32 has been set to 0(using fp16)
outputs coverage
outputs bboxes

If I clean and rebuild with tensorrt disabled then the video plays, but there is no bounding boxes on vehicles.

Any ideas?



hi alex, it takes some time in initializing TensorRt models.

I had same issue as Alex in Jetpack 3.3 & it does indeed take a while for the video to come up. For me, on TX2 , almost 4 minutes.

It is normal. Initialzing TRT models takes certain time.