The Tianmouc Vision Chip boasts record-breaking image processing capabilities, reaching up to 10,000 frames per second.
Scientists from Tsinghua University in Beijing, China, have announced the development of the world’s first brain-inspired vision chip, equipping machines with human-like visual perception, as reported by Interesting Engineering on May 31. The research has been published in the journal Nature.
Tianmouc Vision Chip. (Photo: Tsinghua University).
AI Visual Perception is laying the groundwork for a groundbreaking technological revolution, especially in automated systems like self-driving cars. However, achieving effective, accurate, and flexible visual perception in diverse, dynamic, and unpredictable environments remains a significant challenge.
The new chip named Tianmouc is touted as the fastest vision chip in the world, featuring record image processing speeds. It captures visual information at rates of up to 10,000 frames per second, with a 10-bit accuracy and a dynamic range of 130 decibels. Additionally, Tianmouc reduces bandwidth usage by 90% while maintaining low power consumption.
Tianmouc is inspired by the human visual system. It analyzes visual information through two pathways: one for perception and another for rapid response.
“This is a perceptual chip, not a computational chip, based on our original engineering roadmap. First, it balances speed and dynamic performance within the vision chip, while introducing a computational method that differs from existing machine vision strategies. Second, this approach mimics the dual pathways of the human visual system, allowing for effective decision-making,” stated Shi Luping, project leader and professor at the Brain-Inspired Computing Research Center (CBICR) at Tsinghua University.
The research team asserts that Tianmouc will open new avenues for advancements in autonomous driving, defense, and could lead to the development of many new applications.
