The Popularity of Webinar Platforms and Online Meeting Software
The popularity of webinar platforms and online meeting software has gone through the roof in recent months, with many companies and educational institutions going remote and online.
Real-time Object Detection in Live Video
Real-time object detection can enhance the user experience during a live stream. In a nutshell, object detection is used to detect instances of semantic objects of a certain class (like people, buildings, or cars) in digital images and videos. Today, it has multiple business applications, such as image search, facial recognition, and autonomous driving, etc.
People Use Webinars and Online Meetings More than Ever Before
Because of the COVID-19 pandemic, many of us have been stuck at home, working and studying remotely, and meeting online, not being able to visit the office or school. Existing software providers who sell webinar and online meeting platforms have hugely benefited from this. According to MarketWatch, Zoom’s daily active user count in March 2020 was up 378% from the previous year, while Microsoft reported a 775% bump in the use of its cloud applications overall, due to the surge in the volume of remote work and online learning.
Ways to Use Real-time Object Detection in Live Video
The challenge for software dealing with video conferences, webinars, or live streams is detecting and classifying objects in real-time – while also maintaining high performance, which is crucial for the productivity and efficiency of online meetings.
Replacing the Background with a Selected Image, Video, or GIF
This has become all the rage recently, when more teams around the world started working remotely and needed a way to make countless online meetings more entertaining. Meeting participants can select an available photo background or choose one from their desktop folder or phone camera roll. The software detects the face, neatly placing it against the background, preferably without impacting the overall meeting performance too much.
Blurring the Background
This helps to focus the image on just the person in the foreground. It’s again a very popular feature introduced recently, e.g., by Skype, allowing people to be more comfortable during business calls with video, not paying attention to what (or who) is behind them.
How to Build Your Own Image Recognition System
We’ve recently published an ebook that shows precisely what’s needed to build leading-edge software using object recognition algorithms. Download it and find out, step by step, how to:
- Define project scope and metrics
- Collect data and use synthetic data
- Train models and test their performance
- Deploy the models to production
- Monitor and optimize their performance over time
And if you’re looking for expert advice on using object detection in your software – chat with a DLabs AI specialist today.
Conclusion
Real-time object detection can enhance the user experience during a live stream. It’s a popular feature introduced recently, allowing people to be more comfortable during business calls with video, not paying attention to what (or who) is behind them. By using real-time object detection, software can detect and classify objects in real-time, allowing for more efficient and productive online meetings.
Frequently Asked Questions
- What is real-time object detection in live video?
Real-time object detection is used to detect instances of semantic objects of a certain class (like people, buildings, or cars) in digital images and videos. - How does real-time object detection enhance the user experience during a live stream?
Real-time object detection can enhance the user experience during a live stream by allowing for more efficient and productive online meetings, and more. - How do I build my own image recognition system?
We’ve recently published an ebook that shows precisely what’s needed to build leading-edge software using object recognition algorithms. Download it and find out, step by step, how to define project scope and metrics, collect data and use synthetic data, train models and test their performance, deploy the models to production, and monitor and optimize their performance over time.
