Artificial intelligence can already predict many things. It also cannot hide where we will go or what we will do next.
As we can learn from the publication on May 31 scientific work, research from Carnegie Mellon University, Google AI and Stanford University have verified that Deep learning systems are able to predict our actions. How is this possible?
Predicting movement path and behavior
As algorithms and technology develop, systems are able to analyze more and more information, including video. This can be helpful, for example, in applications that help predict pedestrian behavior and avoid accidents. From this perspective, it is therefore extremely important to predict the future path of pedestrians, called future person path/trajectory prediction.
source: youtube.com
As we learn from the scientific work, first the person in the video material is coded with rich visual features, which allow us to capture human behavior and interactions with their environment.
Then they are added auxiliary tasks predicting the location of activities to facilitate the joint training process.
The resulting model the scientists showed the effectiveness of this solution both on popular and recent large-scale video benchmark. Furthermore, it has been shown quantitatively and qualitatively that the next model successfully predicts significant future actions.
source: youtube.com
The analysis of pedestrian activities and the use of rich information functions therefore enable teaching systems to predict future actions and their locations.
Modeling behavior and action paths
The tests carried out by the scientists prove that this method is able to not only guess the future path of movement, but also to produce a significant forecast of future activity. The result is the first empirical evidence that Joint modeling of paths and actions benefits future path predictions.
Use of technology
This is already considered a key element in understanding video. This way of looking at visual information from the past, as scientists declare, can help predict future actions in many applications used, for example, in autonomous cars or robotics.
What's more, it can support public safety, traffic monitoring and manage it by automatically detecting streaming video activity.
Ethical aspects of this solution
The prospect of increasing road safety and solving the problem of autonomous cars to some extent sounds incredibly promising. optimistically. Unfortunately, the solution also has its dark sides.
Future studies of this type on path and activity prediction may raise ethical issues regarding privacy and security. This method has also not yet been tested for different populations, and as we know, people can behave differently in different situations, so predicting their activity should also be adapted to individual populations and their specifics.
Talk to us about AI
Join the "AI Business" group on Facebook
We recommend an e-book about AI
AI in Marketing – How to Increase Sales and Customer Engagement?
The Turing Test: Is AI Already More Intelligent Than Humans?