ETH Zurich and EPFL Unveil AI Camera That Rebuilds Particle Tracks From as Few as Five Photons

Researchers in Switzerland have built a prototype detector that reconstructs the three-dimensional paths of subatomic particles using only a handful of photons — as few as five — by pairing a light-field camera with a Transformer neural network. The system, dubbed PLATON (PLenoptic imAge of Tracked photONs), was developed by teams at ETH Zurich and EPFL and described in a paper published in *Nature Communications*, offering a potential route to dramatically simpler particle detectors and, eventually, medical imaging tools such as PET scanners.