Machine vision is becoming vital for technologies like smartphones, drones, and self-driving cars. However, processing the huge amount of visual data these systems produce requires a lot of power, storage, and computing resources. This makes it hard to use advanced vision systems in small devices.
The human visual system works differently, filtering information to process only what’s needed, using very little energy. Neuromorphic computing, which copies how the brain’s neural networks function, offers a way to make machine vision more efficient. Still, challenges remain, like matching human color recognition and removing the need for external power.
Researchers led by Tokyo University of Science have developed a new device that tackles these issues. Described in a paper published in Scientific Reports, the device is an artificial synapse, a component that mimics brain connections, and can distinguish colors with high precision. It uses two dye-sensitized solar cells, which generate electricity from light, eliminating the need for external power. This self-powered feature makes it ideal for edge computing, where devices process data locally to save energy.
Testing real-world potential
The device can identify colors with a resolution of 10 nanometers, close to human eye precision, across visible light. It produces different voltages for different colors, like positive for blue and negative for red, allowing it to perform complex tasks that usually need multiple devices. In tests, the researchers used the device in a system called reservoir computing to recognize human movements recorded in red, green, and blue. It achieved 82% accuracy in classifying 18 combinations of colors and movements using just one device, unlike traditional systems that need several components.
This technology could improve many fields. In self-driving cars, it could help recognize traffic lights or signs more efficiently. In healthcare, it could power wearable devices that monitor health with low battery use. For smartphones or virtual reality headsets, it could enhance vision features while saving power. The researchers believe this device will lead to low-power machine vision systems that see the world much like humans do, with uses in sensors for cars, medical devices, and portable gadgets.