Skip to content
This fork of the deep learning guide has been adapted to work with a variety of different inputs USB camera, GigEVision and RTP on the TX1 SoM. This is a a quick demonstrator and example for users of the Abaco Systems rugged Small Form Factor (SFF) TX1 boxed solutions. Please visit out website for more details.
Branch: master
Clone or download
Pull request Compare This branch is 145 commits ahead, 381 commits behind dusty-nv:master.
Latest commit 970eda5 May 2, 2017
Type Name Latest commit message Commit time
Failed to load latest commit information.
abaco Launch script for desktop shortcut Apr 21, 2017
camera Fixed optional dependancy on aravis (only needed for GigEVision camer… Mar 15, 2017
cuda Fixed issues in the Bayer GR8 colour conversion function CUDA function Mar 3, 2017
data Added logo file for rendering onscreen Nov 29, 2016
detectnet-camera Added GigEVision camera support into detectnet. Mar 2, 2017
detectnet-console Bulk checkin. Nov 25, 2016
display Adding GigEVision camera support (requires Aravis) Mar 1, 2017
docs auxiliary documentation Sep 7, 2016
imagenet-camera Fixed issues with webcam Mar 23, 2017
CMakeLists.txt Added desktop installer shortcut Apr 21, 2017 Added SDL dev dependancies Dec 5, 2016
FindSDL2.cmake Bulk checkin. Nov 25, 2016
FindSDL2TTF.cmake Updating changes needed to propogate the correct resolution into gst… Dec 13, 2016 Update May 2, 2017
config.h Changing default camera to /dev/video0 for GVC1000 (dev kit is /dev/v… Apr 21, 2017
detectNet.cpp added FP16/FP32 readout in camera samples Sep 26, 2016
detectNet.h added FP16/FP32 readout in camera samples Sep 26, 2016
imageNet.cpp Tidy up code and startup messages. Mar 2, 2017 updated imageNet recognition Aug 11, 2016
imageNet.h Added overlays and more detailed classifer information Nov 29, 2016
loadImage.cpp optional output image + font overlay Sep 24, 2016
loadImage.h optional output image + font overlay Sep 24, 2016
tensorNet.cpp added FP16/FP32 readout in camera samples Sep 26, 2016
tensorNet.h added FP16/FP32 readout in camera samples Sep 26, 2016

Abaco stripe

Abaco Systems TX2 enabled Small Form Factor solutions

Deep Learning for Rugged Environments (Industrial / Military)

This is a branch of the NVIDIA's deep learning inference workshop and end-to-end realtime object recognition library for Jetson TX1 and TX2. Please reffer to the main branch for details and tips on setting up this demonstartion.

Included in this repo are resources for efficiently deploying neural networks using NVIDIA TensorRT.

Vision primitives, such as imageNet for image recognition and detectNet for object localization, inherit from the shared tensorNet object. Examples are provided for streaming from live camera feed and processing images from disk.

Project 'Elroy'

note Project 'Elroy' is a rugged TX2 box built to military specifications. Due for release mid 2017.

ImageNet for more information on source images goto (

Support for USB webcam and GiGEVision

On the Abaco Systems SFF box there is no CSI camera so gstCamera has been modified to support the Logitech C920 Webcam. At the moment only imagenet-camera.cpp has been tested and proven to work with this modification.

Modified pipelines can be supplied to the create function to support cameras using RGB colour space. The included pipeline in imagenet-camera.cpp for the C920 is shown below:

video/x-raw, width=(int)640, height=(int)480, format=RGB ! videoconvert ! video/x-raw, format=RGB ! videoconvert ! appsink name=mysink

Logitech C920 HD USB webcam:

Logitech C920 Webcamera

note : Set VIDEO_SRC to GST_V4L_SCR to use your RGB webcamera as an input source.

ptGrey GigEVision Blackfly camera:

ptGrey Blackfly

note : Set VIDEO_SRC to GST_GV_STREAM_SCR to use GigEVision cameras.

Set VIDEO_SRC to VIDEO_NV revert to the Jetson evaluation on board CSI camera.


The aim of this project is to create a network trained on images that come from military applications such at Air / Sea / Land. Updating the network to work with this updated network for demonstration and as an example to a defence audiance.

Please be patient whilst we build our DIGITS server to retrain the network and watch this space.

  • imagenet-camera.cpp updated to use webcam
  • detectnet-camera.cpp updated to use webcam
  • parametrize requested USB resolution (Default fixed at 1280 x 720)
  • Add suport for GigE Vision Cameras using the Aravis libaraies.
  • GigEVision RGB8 colorspace support
  • GigEVision YUV422 colorspace support
  • GigEVision Bayer8 colorspace support.
  • update training data (military images)
  • update GUI to run in window and toggle to fullscreen (fill screen)
  • update GUI to use SDL2
  • update GUI fonts to SDL2_ttf
  • add RTP video streaming source for YUV encoded video streams. Injest progressive GVA (Generic Vehicle Architecture) compliant video sources.
  • Code tidys up, remove redundant depedencies.

Building from Source

Provided along with this repo are TensorRT-enabled examples of running Googlenet/Alexnet on live camera feed for image recognition, and pedestrian detection networks with localization capabilities (i.e. that provide bounding boxes).

The latest source can be obtained from GitHub and compiled onboard Jetson TX2.

note: this branch is verified against :

  • TX1 JetPack 2.3 / L4T R24.2 aarch64 (Ubuntu 16.04)
  • TX1 JetPack 2.3.1 / L4T R24.2.1 aarch64 (Ubuntu 16.04)
  • TX2 JetPack 3.0 / L4T R27.1 aarch64 (Ubuntu 16.04 kernel 4.4)

1. Cloning the repo

To obtain the repository, navigate to a folder of your choosing on the Jetson. First, make sure git and cmake are installed locally:

sudo apt-get install git cmake

Then clone the jetson-inference repo:

git clone

2. Configuring

When cmake is run, a special pre-installation script ( is run and will automatically install any dependencies.

mkdir build
cd build
cmake ../

3. Compiling

Make sure you are still in the jetson-inference/build directory, created above in step #2.

cd jetson-inference/build			# omit if pwd is already /build from above

For jetson 64bit builds the architecte will be aarch64, with the following directory structure:

   \aarch64		    (64-bit)
      \bin			where the sample binaries are built to
      \include		where the headers reside
      \lib			where the libraries are build to

4. GigEVision additional steps (Optional step for GigE support)

For GigEVision camera support please download and compile the Aravis 0.6 release.

git clone
cd aravis
sudo apt-get install intltool gtk-doc-tools
./ --enable-viewer

Once built do

make install

You will also need the bayer plugin for gstreamer to enable display Once built do

sudo apt-get install gstreamer1.0-plugins-bad

Header files and library objects should not allow you to compile the code with GigEVision support. Configure camera in config.h


Webcam Input

Mug Classified Mug and Mouse Classified

GigEVision Camera Input

Banana Classified


The font file and binary images used can be found in /data and should be copied into the working directort.

  • q = Quit application
  • f = Toggle fullscreen
  • SPACE = Toggle overlay (imagenet-camera only)

Notes on Ethernet jitter (GigEVision)

For testing (on the Jetson TX2) I selected an Intel PCIe Ethernet NIC device that has deeper buffers and can allows pause frames to be disabled. To optimise your network interface device please run the script jetson-ethernet found in the jetson-scripts project under Abaco Systems. Example streams using external NIC:

  • RGB8 encoded video streams at 1280x720@30Htz, total bandwidth consumed is aproximatly 82.2 Mb / Sec. SampleOutput-RGB8.png best for quality.
  • YUV422 encoded video streams at 1280x720@30Htz, total bandwidth consumed is aproximatly 52.4 Mb / Sec. SampleOutput-YUV422.png
  • BAYER GR8 encoded video streams at 1280x720@30Htz, total bandwidth consumed is aproximatly 27.4 Mb / Sec. SampleOutput-Bayer_GR8.png lowest bandwidth. Can achieve 1920x1080@30Htz 61.0 Mb / Sec.

The sample output supplied is a test card encoded using each of the three color spaces. Bayer has a few different formats we are only testing GR8 (requires CUDA kernel for each to convert video).

note : The YUV422 and Bayer GR8 CUDA convesion functions are not optimal and could be improved!!.


Abaco stripe

You can’t perform that action at this time.