Recently, novel concepts and applications of nanophotonics have been demonstrated in many new paradigms. For example, metasurface integrated miniaturized spectrometers can encode and decode high-dimensional spectral information assisted by deep learning tools (Science 371, eabe0722 (2021); Nanophotonics 11, 2507-2529 (2022)). Beyond computational spectroscopy, deep learning algorithms can map the nonlinear relationship between unknown light beam property to corresponding photoresponses, thus ideally utilized for computational sensing for detecting high dimensional spectral information (Science 379, eade1220 (2023)). Moreover, photonics are a class of passive elements that showed neuromorphic computing capability with high accuracy and low optical losses (Nature Photonics 15, 102-114 (2021)). The integration of metasurface and AI algorithms with many popular two-dimensional semiconductors has also powered high-performance optoelectronic devices such as machine vision realized through reconfigurable sensor array, polarimetry and wavelength detection through deep neural network (Nature 579, 62-66 (2020); Nature 604, 266-272 (2022)). This session will gather experts who work closely with nanophotonics, low-dimensional optoelectronics and AI algorithms that can promote advanced interdisciplinary applications for computing, sensing and imaging techniques.