Key Takeaways
1. The AI program “Jack of Many Faces” identifies drunk, tired, or angry drivers by analyzing their facial movements and expressions in video footage.
2. The system boasts high accuracy rates, recognizing blood alcohol levels with 90% accuracy and drowsiness with 95% accuracy.
3. It categorizes drivers into sober, moderate, or severe states, providing a comprehensive assessment of driver safety.
4. Unlike traditional methods, this technology operates passively and continuously in real-time without requiring physical interaction from the driver.
5. The system utilizes a combination of color video and infrared night-vision footage to enhance performance in low-light conditions.
Researchers at Edith Cowan University (ECU) in Australia have developed an artificial intelligence program that can identify drivers who are drunk, tired, or angry just by looking at video footage of their faces. This program, called “Jack of Many Faces,” assesses small facial movements, eye blinking, and general expressions to keep an eye on three key reasons for road accidents.
High Accuracy Levels
The team behind the research claims that this system can recognize blood alcohol levels with almost 90% accuracy, while it detects drowsiness with a 95% accuracy rate. It can also classify drivers into three different categories: sober, moderate, or severe. Since high fatigue can sometimes look similar to being drunk, and anger can lead to road rage, observing all three emotional and physical states gives a thorough evaluation of driver safety.
Understanding Driver States
This algorithm is clever, as it can determine whether a driver is tired, just showing a facial expression, or under the influence of alcohol. By distinguishing between these elements, it gains a better insight into the driver’s true physical condition. — Zulqarnain Gilani, a researcher involved in the project, said.
Unlike traditional breathalyzers and blood tests, which are very accurate but invasive and require cooperation, this new technology works passively and continuously in real-time without needing any physical interaction from the driver.
Night Vision Enhancements
To make sure the technology is effective during the night, the research team created a companion model that smartly combines regular color video with infrared night-vision footage. By fusing these two different types of video streams, the system can accurately capture essential facial features in low light, enhancing its performance.
ECU’s innovative approach could change the way we think about road safety and driver assessments.
Source:
Link
Leave a Reply