We Use Cookies
This website uses cookies to ensure you get the best experience on our website.
annotation of facial features using object detection and classification
For a leading In-Vehicle Infotainment (IVI) design company providing electronics solutions for automotive customers worldwide
0
M+
Number of Transactions
0
%
Accuracy Level
0
%
Productivity Increase
About the Customer
This client is a leading In-Vehicle Infotainment (IVI) design company providing electronics solutions for automotive customers worldwide. They provide inventive and advanced embedded, Multimedia, and Connectivity solutions to 2 out of the top 3 automobile manufacturers in Europe, US, and China along with turnkey services and certification support. A make in India initiative, they have a global presence in the US, China, Japan, Taiwan and South Korea.
Problem
To become the leader in infotainment and automotive solutions, the client realised that Driver Monitoring was emerging as an essential requirement for Advanced Driver Assistant Systems (ADAS) and autonomous driving and hence had to develop driver safety modules that includes Face Recognition, Emotion Detection, Child detection and Activity detection.
While Driver Management Solutions have been developed using computer vision and other image processing approaches, fine tuning of parameters like specific facial features and annotating head and eye tracking and blink rate monitoring was required.
Approach
NextWealth, as a pioneer and expert in annotation and computer vision, partnered with the client to prepare data for their Machine Learning (ML) models. NextWealth spearheaded the annotation of facial features using object detection and classification. With Head Pose detection and gaze detection being a challenge for the client, NextWealth stepped in to identify and tag yaw, pitch and roll along with other environmental factors such as lighting, occlusions and expressions to develop a person-independent, non-intrusive model and algorithm.
Results
By integrating and incorporating gaze detection and head pose estimation in driver safety modules, algorithm accuracy was improvised, and the initial prototype was successfully deployed.