Accelerating Large-Scale Object Detection with TensorRT

Originally published at:

Detecting the presence of humans accurately is critical to a variety of applications, ranging from medical monitoring in nursing homes to large-scale video analytics in various environments. High performance for deep learning training makes it possible to create robust and generalizable models for objects, humans, animals, and machines. Maintaining real-time inference performance in production environments…

Is there a way to implement Yolov3 (maybe yolov3-tiny) with TensorRT?

Sorry for late reply. Definitely you could! All you need is to implement neural-net layers which are not supported in TensorRT(v5 now) as custom plug-in layers.

when can I use this on my phone?

It could run on mobile phone, which just not full real-time for now. And this work is mainly focus on scale-out for large-scale application on GPU server side.

@shounanan:disqus : Where can I find the code to try this out? Thanks

Nice Article. And also can you tell me what is the Precision of YOLOV2 Model ( 100 FPS YOLOV2) . It didn't mentioned anywhere.

Sir, I want the customise the faster rcnn with the tensorrt plugin. I have read from the documentation that faster rcnn with the object detection with tensorrt plugin works with the fixed size 3 channel 375x500 images as input. When I input image of size 5000 *600 faster rcnn does not gave any results. I am struck with this and I don't know how to cope with this error.

Any ideas are welcomed here.