StradVision has developed a software solution that uses deep learning to recognize pedestrians and even texts on road signs. Established in 2014, the ICT firm is armed with experts specializing in software—the majority of its employees have Master’s or doctoral degrees.
The performance of the deep learning-based image detection technology developed by StradVision is regarded as top-notch in the field of front detection camera. The deep learning-based object detection technology is excellent at accurately recognizing not only distant objects but also overlapped ones; it can distinguish one object from another—for instance, between vehicles and pedestrians or between pedestrians and bicycles—and utilizes that information as data for safe driving.
It can also read the movement patterns of vehicles and the posture of pedestrians and predict a situation in advance. StradVision is also deemed to be superior to other domestic and foreign firms utilizing deep-learning image detection technology in terms of a high-efficiency algorithm optimized for vehicles.
StradVision possesses 14 patents, including a method for rapidly reading texts embedded in images, fast parallel data processing and detection accuracy enhancement technology, with 10 of them having been registered in the U.S.
As an automotive supplier with skilled manpower and know-how, Hyundai Mobis is expected to make great strides in deep learning camera detection through technical synergy with StradVision.
With this investment, the firm aims to secure its own technology in camera following its achievement in radar. Paying attention to deep learning, which is the core of AI technology, the firm will strive to gain a competitive edge in sensor technology for self-driving vehicle.
The deep learning-based image detection technology is attracting the spotlight especially as the machine learns accumulated data and finds answers by itself. Recently, many companies are competing for developing image detection technology based on deep learning. No one has succeeded in commercializing this technology as of yet, and it is expected to take 4 to 5 years until its commercialization. Hyundai Mobis plans to develop a deep learning-based camera by 2020, thereby securing unmatched competitiveness in sensor technology for self-driving.