A new type of artificial eye , conceived by combining electronic light-sensing devices with a neural network on a single tiny chip, is capable of recognizing what it is seeing in just a few nanoseconds . It is a much faster time lapse than existing image sensors so far.
Object detection is the part of computer vision that studies how to detect the presence of objects in an image based on their visual appearance.
Design that mimics nature
The design of this chip, published in the journal Nature by researchers at the Institute of Photonics in Vienna, Austria, mimics the way that the eyes of animals pre-process visual information before passing it to the brain .
That is, for image recognition to be much faster and much less energy is used in it, a sensor captures and processes the image at the same time . Generally, two parts can be distinguished in the detection process: the extraction of characteristics from the content of an image and the search for objects based on those characteristics. In this new research, however, the data does not have to be read and processed by a computer, but rather the chip itself provides information about what you are seeing.
The team of researchers built the chip from a sheet of tungsten diselenide a few atoms thick, etched with light-sensing diodes. The diodes were then connected to form an artificial neural network, a computational model loosely inspired by the behavior observed in its biological counterpart.
The material used to make the chip gives it unique electrical properties so that the photosensitivity of the diodes, the nodes of the network, can be modified externally. This means that the network can be trained to classify visual information by adjusting the sensitivity of the diodes until it provides the correct answers.
In the study published in Nature , it is demonstrated how the chip was trained to recognize stylized and pixelated versions of the letters “n”, “v” and “z” .
This new sensor is another step on the way to advance the hardware in which an artificial intelligence (AI) is installed, making it faster and more efficient. All in all, there is still a long way to go. For starters, the eye consists of only 27 detectors and cannot control much more than 3×3 block images .
“Our test chip is still small but the technology can be easily expanded depending on the task you want to solve,” says Thomas Mueller , leader of the research. “In principle, the chip could also be trained to distinguish apples from bananas, but we see its use more in scientific experiments or other specialized applications.”
Still, small as it may be, the chip can perform a number of standard supervised and unsupervised machine learning tasks, including, as seen, letter classification and encoding. Machine learning techniques aim to automatically differentiate patterns using mathematical algorithms. These techniques are commonly used to classify images.
Reference: Mennel, L., Symonowicz, J., Wachter, S. et al. Ultrafast machine vision with 2D material neural network image sensors. Nature 579, 62-66 (2020). https://doi.org/10.1038/s41586-020-2038-x
