Abstract
Ultrafast temperature field detection and identification is crucial for applications ranging from environmental sensing and biomedical monitoring to thermal management in advanced energy systems. Conventional temperature sensors-comprising discrete sensing arrays, data storage units, and external processors-suffer from high latency due to slow sensor response, repeated analog-to-digital conversions, and extensive data transmission inherent to von Neumann architectures. Here, we report a diamond array-based quantum sensor that integrates temperature sensing and real-time processing within a unified in-sensor computing (ISC) architecture. Exploiting the strong linear correlation between temperature and the zero-field splitting of nitrogen-vacancy (NV) color center centers in diamond, we realize a fixed-frequency temperature sensor with ultrafast response and tunable responsivity, enabled by multi-parameter microwave modulate. Matrix-vector multiplication of temperature intensity and responsivity, combined with Kirchhoff's current summation, enables direct execution of neural-network-style computations on sensed data. The proposed system achieves a single-shot detection and identification latency of just 196.8 μs, as experimentally validated. This work demonstrates a scalable ISC-enabled quantum sensing paradigm, offering a promising route toward high-speed, low-power intelligent temperature field detection.