Trying to load a different video for the car detection demo on the TX2 but i get an error

Robotics & Edge Computing Jetson & Embedded Systems Jetson TX2
1

I’m new to Jetson TX2 but im familiar with deep learning and I wanted to know if I am doing things the correct way.

I ran the sample for detecting cars in the video. I assume that training was done and everytime i loaded it, it was using cached data (weights/model etc) and would re run the demo quicker.

I assumed that now i could just swap the video file and see if it could detect cars in a different scene. I downloaded a CCTV footage video off the internet, saved it to 10fps, 1080p and mp4 using the h264 codec usign Adobe Premiere and I tried loading the video. However it gives me an error that it could not find some nal or EOF

Could anyone please help me with what codec etc i have to use? I create my content on a windows PC using adobe premiere. Here is my log, sample.mp4 is my new video i downloaded and saved. Thanks!

./backend 1 …/…/data/Video/sample.mp4 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
Net has batch_size, channel, net_height, net_width:1 3 540 960
Using cached TRT model
outputDim c 1 w 240 h 132
outputDimsBBOX.c() 4 w 240 h 132
Failed to query video capabilities: Inappropriate ioctl for device
NvMMLiteOpen : Block : BlockType = 261
TVMR: NvMMLiteTVMRDecBlockOpen: 7647: NvMMLiteBlockOpen
NvMMLiteBlockCreate : Block : BlockType = 261
Failed to query video capabilities: Inappropriate ioctl for device
Failed to query video capabilities: Inappropriate ioctl for device
Starting decoder capture loop thread
Could not read nal unit from file. EOF or file corrupted
Could not read nal unit from file. EOF or file corrupted
Input file read complete
TVMR: NvMMLiteTVMRDecDoWork: 6531: NVMMLITE_TVMR: EOS detected
TVMR: TVMRBufferProcessing: 5486: Processing of EOS
TVMR: TVMRBufferProcessing: 5563: Processing of EOS Done
Segmentation fault (core dumped)

2

Hi LeMoth,
The input is h264 stream. You can use ffmpeg or Yamb http://yamb.unite-video.com/ to extract raw h264 from mp4.

3

hi DaneLLL

this worked thank you very much! However if I were to use this from a webcam stream, is there a library to decode the raw h264 in realtime?

thanks!

4

Hi LeMoth,
For using tegra_multimedia_api, you have to integrate by your self. Not sure but live555 should help your case.
http://www.live555.com/liveMedia/
Other users can share their experience.

You can also consider DeepStream SDK for Jetson