In-car AI imaging systems being refined for autonomous driving

The rapid progress in autonomous vehicle technology is being driven by the integration of advanced computer vision technology, which incorporates next-generation sensors, cameras and sometimes biometric algorithms for various purposes such as driver assistance, surveillance, and entertainment. Within the field of in-car imaging systems, a wide range of hardware and software components are at play, and there has been significant growth in the development of semiconductor devices that embed advanced computer vision models at the edge.

Nonetheless, despite the imperative for advanced driver assistance systems (ADAS) through in-car imaging solutions, original equipment manufacturers (OEMs) face the critical task of ensuring the functional safety of their camera-enabled ADAS. They must take measures to safeguard against cybersecurity threats since the malicious manipulation of sensor data could lead to system failures and unauthorized access to internal cabin images.

OEM developers are refining in-car imaging systems, while researchers are exploring innovative approaches to enhance real-time driver monitoring systems.

Omniq adds in-car face detection feature for security purposes

Omniq, a company specializing in providing artificial intelligence based solutions, has introduced an innovative in-car face detection feature capable of identifying and recognizing faces within a vehicle. This advancement is expected to be a valuable asset for automotive manufacturers, as it can be seamlessly integrated into autonomous vehicle systems to enhance security and prevent potential criminal activities.

The Omniq’s AI capabilities are based on the patented neural network algorithms that mimic human pattern recognition abilities, facilitating efficient and intelligent decision-making. As per the reported data, the company has deployed more than 20,000 AI-based machine vision sensors worldwide, showcasing the accuracy through advanced algorithms and machine learning.

“We believe that this new feature will further improve safety and prevent crimes and terror attacks more efficiently and quicker. The momentum in the market acceptance for our AI based solution continues after our recent wins in additional airports and for safety and security” says Shai Lustgarten, chief executive officer at Omniq.

The integration of a face detection mechanism within autonomous vehicles has become increasingly important. This in-car imaging system offers a comprehensive solution to mitigate accidents resulting from driver fatigue, alcohol consumption, and negligence in maintaining proper front-view attention. Omniq has conducted security tests to showcase the performance of its face detection technology compared to competitors in the field, underscoring its effectiveness.

SemiDrive, Kankan Tech collaborate on improving in-car imaging system

SemiDrive and Kankan Tech, a full stack optical image product and solutions provider, have come together to improve the capabilities of in-car imaging solutions.

In this partnership, SemiDrive provides its X9, an intelligent automotive-grade chip that plays a key role in powering advanced automotive systems. Conversely, Kankan Tech will use this chip to offer a complete imaging ecosystem development service at the platform level. This service includes various technical support aspects for customers, such as custom camera module development, production assistance, algorithm integration, image calibration, and the conduct of road tests.

Kankan Tech has previously demonstrated its expertise by developing high-resolution, retractable intelligent cabin cameras, which are deployed for cabin monitoring and the enhancement of driver assistance systems. Additionally, they have created a camera monitoring system that serves as a modern alternative to conventional rearview mirrors, utilizing camera-based technology.

Furthermore, Kankan Tech has introduced a biometric recognition technology based on palm vein authentication and identification, which enhances vehicle access and security features.

KanKan developer Remark Holdings also recently won a contract to equip police cars in Brazil with facial recognition.

Facial landmark-based behavior recognition can enhance driver monitoring

A recently published research paper introduces an innovative real-time monitoring system that leverages advanced technology, incorporating facial landmark estimation and infrared cameras. The primary aim of this research is to enhance driver safety and attentiveness during their journeys by continuously monitoring the driver’s behavior and issuing timely alerts to prevent accidents resulting from inattentiveness and drowsiness.

The facial landmark estimation is used to identify and track specific points on the driver’s face. This extracts key information about the driver’s head posture and the condition of their eyes. The system initiates by detecting the driver’s face through face detection techniques and subsequently extracts additional insights from the detected facial region.

The proposed method uses two modules for behavior recognition – head pose analysis and eye closure recognition. The head pose analysis focuses on analyzing the driver’s head movements along both horizontal and vertical axes. This information can be used to assess the driver’s level of attention. If the system detects unusual head movements indicative of inattention, it can trigger alerts or warnings to the driver. The eye closure recognition is responsible for detecting instances of drowsiness based on the driver’s eye behavior.

The researchers opted for IR cameras because they are not affected by changes in lighting conditions, making them suitable for monitoring drivers in various conditions, including situations like driving through tunnels and during nighttime. The camera is strategically placid on the steering wheel of the vehicle to ensure accurate monitoring of the driver’s face and facial expressions.

The algorithmic process comprises several steps, including real-time face detection utilizing YOLO v7 algorithms on captured IR images. Following face detection, facial landmarks are extracted, representing specific facial points like the eyes, nose, and mouth, which are subsequently used to analyze facial expressions and head orientation.

The ultimate goal of such research work is to provide a solution that can be integrated in commercial deployments. The authors say “we are dedicated to rigorous testing and exploration to pave the way for its market entry.”

About the author

Abhishek Jadhav is a Master’s graduate in Electrical Engineering and a technology and science writer at EdgeIR,

Article Topics

automotive biometrics  |  biometrics  |  biometrics at the edge  |  computer vision  |  KanKan AI  |  Omniq