One of the most interesting areas in which the field of computer vision has been investigated includes the platform for streaming video. Video streaming has emerged as one of the top areas in the past decade particularly with the lockdown which has led to an increase in viewing audience on these platforms than at any time before.
To enhance the user experience Many of these companies use the power of computer vision (CV). In addressing the attendees at CVDC 2020, Anubhav Shrivastava the head of Data Sciences at Voot Viacom18shared some of the ways Voot employs computer vision.
Computer Vision Has Gained Traction Through The Years
Shrivastava explained that the field of computer vision evolved throughout time and has witnessed the growth of applications over time. He described CV as the studying techniques that assist machines in interpreting, seeing and produce visual images.
The most popular applications domains that make use of computer vision include surveillance, biometrics, checkouts, heat maps, facial recognition remote sensing for medical imaging modeling in 3D intelligent image processing, and many more. The methods used in computer vision the most widely used include image detection, object detection semantic segmentation, in-situ segmentation, deep-tracking, GANs, and much more.
Computer Vision At Voot
Voot is an ad-supported streaming service that is subscription-based and accessible in India. It is home to greater than 100 million monthly peak users. https www voot com activate To operate on this number of users, Voot largely depends on the latest technologies, such as machine learning and artificial intelligence and computer vision, which is an essential technology.
Shrivastava revealed his belief that computer vision has an important role in streaming video as it aids in improving users’ experience, improving retention, boosting efficiency through automation, reducing costs and increase the opportunities for monetisation as well as increase the number of users who sign up and many more. It is used at the supply-side for the creation of content, and also on the side of consumption to enhance user experience” the author said.
The areas in which Voot makes use of computer vision include:
Ads trigger point detectorThe video’s content is set with ads between, which play a crucial part in determining the viewers. The ads must be carefully placed since its placement in the wrong spot at the wrong time could result in the audience having to walk away.
The placement of ads in line to the contents is vital. The use of CV has allowed Voot to choose where to put ads. make sure that the experience of watching videos is not compromised, replace manually detected detection, remove human error, deal with scale , and allow to experiment.
Contextual advertisements: Another important aspect is to display ads that are relevant for users based on video objects. Utilizing technologies like facial recognition, object detection as well as color recognition, firafollower Voot is playing on the psychology of humans to imitate human behavior.
They’ve been able dramatically increase the engagement of users as well as the likelihood to purchase and also ensure that they earn more money.
product placementsProduct positioning is an well-known idea within Hollywood and is being utilized in all kinds of videos across the world. The majority of the times the content is developed first , and then they search for buyers or they create the content for TV and transferred online.
In these situations it is often difficult to guarantee ad placements. So product placements are a image where logos of the product are carefully placed in the videos. Computer vision can be used to identify whitespaces, deep tracking logos as well as to ensure the post publishing of advertising and other.
Highlight creationHighlights (or short videos) are in high demand since the majority of viewers want to view highlights of important episodes at a time. While making highlights for sports is straightforward since it permits voice detection when important events occur that involve the crowd cheering or high-pitched commentary, it’s difficult to accomplish this for non-sports-related videos. Shrivastava states that it’s not just about decibels.
It needs to be able to see the most important moments, observe the footage attentively, look at facial expressions, and so on. Voot is a producer of many highlights and is currently experimenting with computer vision to make and expand it. It’s not operational yet, but the experiments are on.
Augmented Reality: Voot also seeking to integrate the use of augmented reality in its shows to boost participation of the viewers. Particularly during the time of COVID the company is making sure that viewers are part of the show even at home. It is also in an experimental stage, and Shrivastava believes it could become a reality within 2 to 5 years.
Shrivastava ended by saying that the main areas in which computer vision is used can be used to enhance the engagement of users, improve user experience, and to increase the amount of money that can be made. The company has employed methods like group normalisation as well as video to video synthesis in order to produce thematic content for the audience in a certain way.