Vision and Sensing

Breakthroughs in Vision & Sensing

Learn about our global initiatives unlocking complex application environments and new business opportunities. See how these projects push the boundaries of vision and sensing.

OUR WORK

Argus: A Compact and Versatile Vision Foundation Model

Argus is a powerful vision foundation model designed to handle 12 different computer vision tasks—from smart cities to retail—on both cloud and edge devices. By training on just a fraction of the usual data, we’ve created a high-performance backbone that is versatile, practical, and significantly more accessible to deploy.

Learn More

OUR WORK

Stretching Each Dollar: Diffusion Training from Scratch on a Micro-Budget

We’ve developed a strategy to democratize high-end generative AI by training a 1.1 billion parameter model for under $2,000—over 100x less than traditional stable diffusion costs. Using smart masking and synthetic data, we achieve high-quality image generation while proving that top-tier research doesn't always require massive computational budgets.

Learn More

OUR WORK

Beyond RGB: Adaptive Parallel Processing for RAW Object Detection

Traditional image processing is built for human eyes, but it often throws away data that AI needs to see clearly. Our Raw Adaptation Module (RAM) mimics the human visual system’s parallel processing to extract more detail directly from raw sensor data, leading to state-of-the-art object detection even in difficult lighting and high-dynamic-range environments.

Learn More

Futuristic server room with a glowing dome featuring circuit patterns and a brain icon

AI Pipeline

Our goal is to maximize imaging and sensing capabilities using AI. Traditional RGB pipelines are optimized for human perception, and often discard information that could be critical for specific applications, such as scientific analysis or autonomous navigation. We are creating AI-optimized pipelines for computer vision.

A person types on a laptop, with digital gears and flow lines emerging

Vision Foundation Models

Our compact vision foundation models will help shape the future of edge AI, especially given the limited memory and computational capabilities of edge devices. These models provide numerous advantages beyond just deployment flexibility. They require fewer compute resources, enable faster inference speeds, and contribute to smaller carbon footprints.