The demand for accurate perception of the physical world leads to a dramatic increase in sensory nodes. However, the transmission of massive and unstructured sensory data from sensors to computing units poses great challenges in terms of power‐efficiency, transmission bandwidth, data storage, time latency, and security. To efficiently process massive sensory data, it is crucial to achieve data compression and structuring at the sensory terminals. In‐sensor computing integrates perception, memory, and processing functions within sensors, enabling sensory terminals to perform data compression and data structuring. Here, vision sensors are adopted as an example and discuss the functions of electronic, optical, and optoelectronic hardware for visual processing. Particularly, hardware implementations of optoelectronic devices for in‐sensor visual processing that can compress and structure multidimensional vision information are examined. The underlying resistive switching mechanisms of volatile/nonvolatile optoelectronic devices and their processing operations are explored. Finally, a perspective on the future development of optoelectronic devices for in‐sensor computing is provided.

Contact resistance ( R _C ) is of great importance for radio frequency (RF) applications of graphene, especially graphene field effect transistors (FETs) with short channel. FETs and transmission line model test structures based on chemical vapor deposition grown graphene are fabricated. The effects of employing traditional lithography solvent (Acetone) and strong solvents for photo resist, such as N, N-Dimethylacetamide (ZDMAC) and N-Methyl pyrrolidone (NMP), are systematically investigated. It was found that ZDMAC and NMP have more proficiency than acetone to remove the photo-resist residues and contaminations attached on graphene surface, enabling clean surface of graphene. However, strong solvents are found to destroy the lattice structure of graphene channel and induce defects in graphene lattice. Clean surface contributes to a significant reduction in the R _C between graphene channel and metal electrode, and the defects introduced on graphene surface underneath metal electrodes also contribute the reduction of R _C . But defects and deformation of lattice will increase the resistance in graphene channel and lead to the compromise of device performance. To address this problem, a mix wet-chemical approach employing both acetone and ZDMAC was developed in our study to realize a 19.07% reduction of R _C , without an unacceptable mass production of defects.

Conventional photodiodes can distinguish subtle illumination variations, but fail to retain information. Conversely, optoelectronic memories can store light information but encounter difficulties in multiple-state storage due to noise from the deexcitation recombination of the photogenerated carriers. Here we develop an optoelectronic memory that can generate a multilevel response to light stimuli and retain the multi-states in a nonvolatile manner. In the designed type-III heterojunction device, electrons flow from high-Fermi-level (E_f) p-type amorphous carbon to low-E_f n-type TiO_x, generating an unusual built-in electric field that enhances majority carrier transport. Photogenerated electrons from the oxygen vacancies in TiO_x quickly combine with holes from amorphous carbon, effectively reducing deexcitation recombination with charged vacancies and associated noise. The resulting device achieves 1,024 distinguishable optoelectronic memory states without any denoising process, enabling high-precision spatiotemporal information encoding. The thousands of states in optoelectronic memory allows for the emulation raptor vision to perception fast-moving objects.