Fig. 13: In-sensor computing with solution-processed materials.

Schematic diagram of the conventional computing architecture (a) and in-sensor computing architecture (b). c–f In-memory light sensing for artificial visual perception. c Schematic representation of the human visual system for sensing, memory, and computing. d Schematic representation of the graphene/MoS_2-xO_x/graphene photomemristor. e The execution of image processing utilizing diverse operators is demonstrated through the manipulation of distinct states and polarities of the photomemristors. f Schematic illustration of the single-layer perceptron photomemristors array for classifier emulation. Photomemristors of the same class (color) are interconnected in parallel to generate the output current for the activation function. g–l Multimodal in-sensor computing with Mxene-ZnO memristor array. g Schematic illustration of the flexible memristive devices. h Cross-sectional TEM image of the as-prepared ITO/MXene-ZnO/Al device. i Photograph of the flexible memristor arrays on the plumeriarubra. j–l Demonstration of real-time training for image recognition using the memristor device under different rate of humidity. Panel c–f reprint with permission from ref. ²⁰⁵. Nature Publishing Group 2023. Panel g–l reprint with permission from ref. ²⁰⁴. Wiley-VCH 2021.