DeepStream: Next-Generation Video Analytics for Smart Cities

Originally published at:

Imagine every parent’s worst nightmare: a child lost in a crowded mall. Now imagine a scenario where that child is located and rescued in a matter of minutes using a network of cameras deployed within the building—and all the video is recorded, retrieved and analyzed in real time. This represents just one of many possibilities offered…

Hello? Thanks for uploading this article. I have downloaded DeepStream SDK and found there is an example of resnet-18 that seemed to be modified to detect object boundaries. Since the resnet-18 of DeepStream SDK is quite different from the original resnet-18, I wish to download train.prototxt for the resnet-18 of DeepStream SDK; the DeepStream SDK isn't shipped with it. Where can I get it? Thanks,

Are these interfaces (DeviceWorker,Models, and Tensors) supported on the DeepStreamSDK for Jetson?

Hi Vivek,
This article relates to DeepStream SDK version 1.5 for Tesla. The SDK applies to the Tesla family of GPUs and not Jetson. Therefore the DeviceWorker (and other classes included in deepstream.h) would not apply to DeepStream for Jetson. However the models and their output tensors processing logic are based on TensorRT, and hence can be used with the Jetson SDK as well.
Please refer to DeepStream 2.0 for Tesla for commonality to the DeepStreamSDK for Jetson. They are both GStreamer based and share much of the interfaces and software architecture. We plan to continue to improve compatibility between our Jetson and dGPU platforms.

Thanks for the response!