Quantum gas microscopy is an in-situ imaging technique used to investigate many-body phenomena in cold-atom quantum simulators and can provide resolution at the single-particle level; however, limiting factors, such as short lattice constants and finite signal-to-noise ratios, weaken image resolution. Here, the authors develop an algorithm based on unsupervised deep learning that can reconstruct the occupation of an optical lattice of Cs atoms from fluorescence images with high fidelity.
- Alexander Impertro
- Julian F. Wienand
- Monika Aidelsburger
