Jetson TX1 Questions

Robotics & Edge Computing Jetson & Embedded Systems Jetson TX1
1

Hi there!

I received my Jetson TX1 couple of days back and I am kind of confused with the whole thing. Can somebody answer some of the questions I have?

  1. I installed the latest L4T and everything went fine. However, I am unable to run the sample executable files from the sample folders. It doesn’t show anything.

  2. I followed some guide to create custom resolution for Ubuntu and for some reason it throws an error every time I do that. Is there a proper guide to do it for Jetson TX1 boards?

  3. I am trying to create an image classifier for an industrial application (my final graduate project). Can somebody recommend a way to start it. I am only familiar with Tensorflow and I’m sure Jetson TX1 has something better than that to do image classification. Recommend me a way to start.

Thanks in advance.

Stanley

2

varughstan,

  1. I installed the latest L4T and everything went fine. However, I am unable to run the sample executable files from the sample folders. It doesn’t show anything.
    => when I tries.
    => ./camera_jpeg_capture
    => it will capture 30 JPEG files, output0xx.jpg, and you should see camera preview screen on the HDMI display briefly.

  2. I followed some guide to create custom resolution for Ubuntu and for some reason it throws an error every time I do that. Is there a proper guide to do it for Jetson TX1 boards?
    => When you said ‘custom resolution’ is for HDMI display to run the intended resolution you want? How did you do that and is there a log output?

  3. I am trying to create an image classifier for an industrial application (my final graduate project). Can somebody recommend a way to start it. I am only familiar with Tensorflow and I’m sure Jetson TX1 has something better than that to do image classification. Recommend me a way to start.
    => NV has an excellent link for you to get started,
    https://www.nvidia.com/en-us/deep-learning-ai/developer/#learn-how

For image classifier, GoogleNet is a good option. mnist is a good image classifier datasets you could get started with. You can also use your own images but you will need to do labelling. From above link, NVIDIA provides DIGITS that has an easy interface for you to play with mnist data set and get familiar with the flow. TensorFlow is a framework and I had not used it a lot yet. Good luck!

3

Hi Chijen,

Thank you for the reply.

  1. It is solved.
  2. Yes, native output resolution for my screen connected via HDMI-0

Here is how I tried it.

ubuntu@tegra-ubuntu:~$ cvt 1440 900 60
# 1440x900 59.89 Hz (CVT 1.30MA) hsync: 55.93 kHz; pclk: 106.50 MHz
Modeline "1440x900_60.00"  106.50  1440 1528 1672 1904  900 903 909 934 -hsync +vsync
ubuntu@tegra-ubuntu:~$ sudo xrandr --newmode "1440x900_60.00"  106.50  1440 1528 1672 1904  900 903 909 934 -hsync +vsync
[sudo] password for ubuntu: 
ubuntu@tegra-ubuntu:~$ xrandr --addmode HDMI-0 1440x900_60.00X Error of failed request:  BadMatch (invalid parameter attributes)
  Major opcode of failed request:  140 (RANDR)
  Minor opcode of failed request:  18 (RRAddOutputMode)
  Serial number of failed request:  20
  Current serial number in output stream:  21
ubuntu@tegra-ubuntu:~$
  1. I will take a look at that. Is OpenCV a good option?

Thanks in advance.

–Stanley

4

Stanley,
Did you use $xrandr to list out all allowed resolution first?

OpenCV has bottleneck but could be handy or useful for certain computer vision task. Some of the tutorials will give you more insight on this.

5 
ubuntu@tegra-ubuntu:~$ xrandr
Screen 0: minimum 8 x 8, current 1280 x 720, maximum 16384 x 16384
HDMI-0 connected primary 1280x720+0+0 (normal left inverted right x axis y axis) 410mm x 260mm
   1280x720      60.00*+  59.94  
   1920x1080     60.00    59.95    30.00    29.97    61.25    60.05  
   1440x480      62.69    60.05  
   1024x768      60.01  
   800x600       60.32    56.25  
   720x480       59.94  
   640x480       59.94    59.94  
  1440x900_60.00 (0x1b7) 106.500MHz -HSync +VSync
        h: width  1440 start 1528 end 1672 total 1904 skew    0 clock  55.93KHz
        v: height  900 start  903 end  909 total  934           clock  59.89Hz
ubuntu@tegra-ubuntu:~$

This is my output for xrandr.