With the widespread application of infrared detection technology in fields such as military reconnaissance, aerospace monitoring, and security early warning, infrared measurement systems play a critical role in infrared detection. In response to issues such as low calibration efficiency and significant environmental interference in the calibration and radiative property inversion of infrared measurement systems, this paper proposes a calibration and radiative property inversion method based on infrared weak small targets. A small-area blackbody source is used as a controllable radiation source to project infrared targets, and deep learning networks are employed for precise identification and gray-scale extraction of infrared weak small targets. Using this, a calibration model for the measurement system is established. Experimental results show that the method demonstrates good calibration stability within the temperature range of 298 K-308 K, with the absolute error of radiative property inversion controlled within ±2 K and the relative error of inversion temperature ≤ 0.5%. Regression analysis also indicates high temperature inversion accuracy (R2>0.94). Compared to traditional methods, the proposed method balances calibration efficiency and accuracy while extending the ability to invert the temperature field of targets. This research provides an effective solution for rapid calibration and high-precision radiative property analysis of infrared weak small targets.