Seminar series focused on NVIDIA "software"

Developers and companies involved in IVA (image analysis) and robot application/model development have their own rooted software development know-how. Although the use of NVIDIA GPUs is increasing and they are aware of the existence of development support software provided by NVIDIA, many people say that they feel that porting their application/model development methods is a very high hurdle.
　

As a Macnica engineer who has seen many customers develop without using NVIDIA software, he understands the ``difficulty of using tools that have never been used before,'' and the ``good points of the NVIDIA SDK/Platform'' that should still be utilized.

"What will be useful in our development?" "Which part of the development flow can be made more efficient?"

In order to answer the question of

First of all, from the super introductory overview that you can feel "I see" why you should use it,

The experience of our engineers who struggled to understand and use NVIDIA software just like our customers

was condensed into this seminar.

Please take a look at the on-demand video at the end of the article.

Learn DeepStream SDK, Isaac, and TAO Toolkit in 4 weeks of continuous intensive lectures

The NVIDIA AI/deep learning development support software delivered over four weeks is an introductory part (1st week: NVIDIA DeepStream SDK, 2nd week: NVIDIA Isaac™, 3rd week: NVIDIA TAO Toolkit), and the second half is practical. (Week 4: DeepStream SDK Practice) so that you can choose the week you are interested in.

agenda

Part 1: DeepStream SDK to accelerate IVA image analysis application development

Information on image analysis application development is scattered, but few are useful, and beginners of DeepStream SDK need to read and understand the documentation. We introduced the plug-ins provided by GStreamer and DeepStream SDK that you should keep in mind, as well as the fact that the series of video analysis processes are modularized in the DeepStream SDK.

Part 2: Isaac Realizing the Digital Twin of Robots

The key to the development of AI robots, which are widely used in the manufacturing industry and logistics, is the work of generating learning models that assume various use cases. It is possible to simulate in virtual space with Isaac before evaluation on the actual machine, and it is possible to overcome the problem that physical data collection is difficult. He also explained the necessity of SLAM technology required for mobile robots, and at the end saw a demonstration of Isaac Sim/ROS working together to operate Visual SLAM.

Part 3 Simplify and accelerate AI model development using TAO Toolkit

He explained how the TAO Toolkit, which is active in the "learning phase" of the AI workflow, can accelerate model development through transfer learning, and how it is linked with the DeepStream SDK in the subsequent inference phase. We also saw demonstrations of hand gesture recognition and speech recognition using the NVIDIA® Jetson AGX Orin™ developer kit and TAO Toolkit, and many people asked us questions about the construction.

4th DeepStream SDK Practice

Mr. Wakasa, CTO of Ajira Co., Ltd., which provides an AI security system that utilizes action recognition AI technology, took the stage and spoke frankly about the use of DeepStream SDK in an actual system. In large-scale streaming, there are difficulties in development and construction, such as handling different camera inputs and linking with external systems. Thank you.

In particular, we received a lot of feedback from people who said, "I want to hear more about this type of case" and "The content was inspiring to see how it could be used." I'm here.

Easy registration to watch on-demand videos

The online seminar introduced in this event report is available for on-demand delivery. We will send you a viewing URL when you register using the form below.