Mit Ai Model Speeds Up High-resolution Computer Vision For Autonomous Vehicles

MIT and MIT-IBM Watson AI Lab researchers have developed a new computer vision model, EfficientViT, which is capable of performing semantic segmentation in high-resolution images in real-time on devices with limited hardware resources. This model is more efficient than state-of-the-art models as it has linear computational complexity and hardware-efficient operations. This model could be used to improve image quality in video streaming or help autonomous vehicles identify road hazards in real-time.