My Experience into Nvidia Jetson Applications

During my early years of employment working at e-con Systems, I was given an opportunity to work on Nvidia Jetson Xavier processor and my task was mainly involved in building camera sensor drivers for new sensors on the Nvidia Jetson platform. My mentor Dilip Kumar and I started exploring various frameworks that are supported by Nvidia(out of our own curiosity to learn something new) and that is when we came across frameworks like Visionworks, LibArgus and Tegra Multimedia API.

What happened next?

We found a common interest and started exploring something apart from just developing drivers for the platform. The next thing we did was… spend a lot of off hours trying to jot down everything that’s available on the platform, attending sessions and training courses on the platform.

Somewhere down the line, I believed that if our company makes camera drivers and sample applications to our customers, why don’t we build our own application that is fully customizable and helps the developer community as a whole.

Development of my first Application:

I started exploring sample applications that were already present for the ‘LibArgus‘ and ‘Tegra Multimedia API‘ framework. Few months later, I created my first Application and named it : ‘eCAM_Argus_Multicamera’ which, till date, I consider as my biggest achievement at the firm. It was initially integrated into the product : ‘SurveilsQUAD‘.

Features supported:

• Dynamically stream from multiple cameras (1-6) on a single application instance
• Synchronously/asynchronously capture images from all actively streaming cameras
• Burst record from all actively streaming cameras
• Support for obtaining and displaying frame statistics such as ISO Sensitivity, Scene Luminance, Exposure time, Sensor Timestamp, etc.
• Modularized application source for easy integration with other frameworks such as OpenCV, VisionWorks, Gstreamer, etc.
• Support for individual camera and ISP control configuration.

I got a chance to publish my first article in the company : Detailed Guide to LibArgus with SurveilsQUAD

Who can be benefited with the guide?

This article can be helpful to developers / students / research grads who are working on:

• Nvidia Jetson platforms – Nvidia Jetson TX1/TX2/Xavier/Nano
• LibArgus framework on C++.
• Tegra Multimedia API.
• Want to integrate camera application with frameworks such as OpenCV, VisionWorks, Gstreamer, CUDA, OpenGL,etc.

I hope people reading this blog find it helpful for the development activities they are involved in. You can reach out to me on my LinkedIn profile for any kind of discussions which I would love to be a part of.

Recognition:

The next few months at the office were like the stream of golden opportunities. The Application was a hit, I was recognized by everyone in the team and was the first person anyone would contact including clients and developers working with the company. I was called for client calls / developer K.T. sessions and review activity for anyone who was working on ‘LibArgus‘ or ‘Tegra Multimedia API‘.

I participated in a Hackathon contest with my fellow colleagues and proposed a new method for streaming ‘monochrome stereo‘ cameras with custom HDR algorithm using ‘LibArgus‘

The knowledge that I had gathered helped us bag the award of being ‘Runner up‘ in the Hackathon Contest.

Two years down the line I was going through the LinkedIn feed and I came across something crazy : ‘AE for Monochrome Sensors‘. The very idea that we proposed during the Hackathon contest got converted into a product solution and I heard has helped earn a virtue to the company. Never been any prouder.