Abstract: Vegetation-based ecological protection has become one of the important methods for slope protection. To study the inhibitory effect of vegetation at different growth stages on the cracking of expansive soil slopes, large-scale expansive soil slope dry-wet cycle tests are carried out. Through digital image processing technology, the growth parameters of vegetation and the morphological characteristics of cracks are monitored in real time during the dry-wet cycle to explore the evolution law of vegetation growth on the crack evolution of expansive soil slopes. Combined with the changes in volumetric water content and matric suction of the slope, the influence mechanism is revealed. The results show that the vegetation growth parameters such as coverage rate and root surface area density increase with the increase of growth time and can be divided into three typical stages: germination period, growth period and mature period. The crack inhibition effect of vegetation has a significant time effect. The crack inhibition effect in the early growth period is limited, and the crack rate of the bare expansive soil slope and the grass-covered expansive soil slope is similar. However, as the vegetation grows, when the coverage rate and root surface area density exceed a certain threshold, the "water retention" effect of stems and leaves and the "bridging" effect of roots become prominent, and the difference in crack ratio between the two increases significantly. Furthermore, based on the transformation of soil pores among the matrix, settlement, and crack domains during the drying process, a predictive model for the crack volumetric fraction that incorporates the effects of wetting-drying cycles and root systems is developed. The relative error (RE) between the predicted and measured values is less than 9%, and the root mean square error (RMSE) is less than 0.36%, verifying the accuracy and reliability of the model. The research results can provide a reference basis for the crack development process and crack volume rate prediction of grass-covered slopes.