In order to breed Pinus elliottii families with excellent growth traits, based on the UAV multi-spectral technology, the genetic variation analysis of the growth traits of different Pinus elliottii families was carried out. Taking 8-year-old P. elliottii forests of 20 half-sib families as the research objects, the tree height and canopy area of 11 months (except for February) in 2021 were quickly extracted by drone multi-spectroscopy, and DBH prediction model was built based on the actual measurement DBH data; the heritability and breeding values of the tree height, crown area and diameter at the breast height were estimated for each lineage of P. elliottii in different months. The results showed that: there was a strong correlation between the predicted DBH value of the established deep learning model based on the canopy area and tree height and the measured DBH true values, where R² was 0.70 and RMSE was 1.83 cm; the heritability of the three growth traits of P. elliottii was between 0.00 and 0.40; a good genetic gain was obtained when the pedigree was selected with a 10% selection rate, and the genetic gain of the three growth traits ranged from 0.21 to 0.79 (the genetic gain of crown area in November was 0.00). According to the breeding values of the crown area and tree height, family selection should be carried out. Finally, families of 1, 6, 8, 9, 10, 16, 18 and 20 should be considered as candidate families. The deep learning model based on the crown area and tree height of P. elliottii could be applied to predict the DBH of slash pine. Three growth traits of slash pine were controlled by moderate heritability. A good genetic gain was obtained with 10% selection