New Skills

New Training: Real-Time Object Detection with YOLOv3 on PyTorch

by Team Nuggets
New Training: Hands-on With AWS Systems Manager Run Command, State Manager, and Session Manager picture: A
Published on March 24, 2021

In this 7-video skill, CBT Nuggets trainer Trevor Sullivan teaches you how to use the YOLOv3 object detection machine learning model to perform inference against still images and network video streams in real-time with NVIDIA GPUs. Watch this new Python training.

Watch the full course: Python Training

This training includes:

  • 7 videos

  • 41 minutes of training

You’ll learn these topics in this skill:

  • Introduction to Object Detection and Computer Vision

  • Setting Up Your Python and Windows Environment

  • Validating NVIDIA CUDA Compatibility with Your GPU

  • Setting Up the YOLOv3 Object Detection Model

  • Running Inference on a Static Image

  • Running Inference on a Video Stream

  • Improving Object Detection Confidence Levels

How to Accelerate Image Detection with NVIDIA GPUs

By harnessing the power of the NVIDIA GPU, YOLOv3 can process object detection against both images and video streams in real-time at high frame rates. In fact, depending on the GPU being utilized, an NVIDIA GPU can increase processing power by 500%. There are a couple of requirements that need to be met before using an NVIDIA GPU with YOLOv3, though.

First, make sure that a compatible NVIDIA GPU is being used. Most modern NVIDIA GPUs are equipped with CUDA cores – the processing core required to accelerate machine learning. The proper drivers also need to be installed and confirmed working properly before other requirements are met.

After the GPU is installed and confirmed working, NVIDIA's CUDA software needs to be installed. The NVIDIA CUDA software is an application layer that sits between the OS and the GPU hardware itself. It is what allows the GPU to be used for machine learning.

After the CUDA application layer is installed, the OpenCV software needs to be installed and configured. The OpenCV library is a computer vision library that YOLOv3 requires. YOLOv3 will not operate without the OpenCV library installed.

Once your environment meets those requirements the YOLOv3 software should work with GPU acceleration.

Recommended Articles