Reconstruction of two-dimensional temperature and CO₂ concentration fields based on the tunable diode laser absorption spectroscopy (TDLAS) and traditional reconstruction algorithm requires multiple line-of-sight measurements in both axial and radial directions for axisymmetric flames. The experimental system is usually complicated, and the reconstruction efficiency is relatively low. Herein, a machine-learning-based reconstruction model is developed and used to simultaneously retrieve the two-dimensional temperature and CO₂ concentration fields from 4.2-μm mid-infrared TDLAS laser absorption measurements for axisymmetric laminar diffusion flames. Compared with the traditional inversion reconstruction method, the machine-learning-based inversion model only needs to scan the central axis of the flame to simultaneously and efficiently reconstruct the two-dimensional temperature and CO₂ concentration field of an axisymmetric laminar diffusion flame, and the model requires less experimental measurements only in the axial direction, which considerably simplifies the measurement system and improves the reconstruction performance.