Unmanned Aerial Vehicle-Structure from Motion (UAV-SfM) is a labor-saving method for measuring plant height and volume. Recent studies have explored yield estimation in various crops using plant height derived from UAV-SfM. This study examined the application of UAV-SfM in grass breeding, focusing on yield estimation for productivity testing trials and evaluating its potential for individual plant selection. Yield estimation in Lolium perenne was analyzed using Random Forest (RF) regression with UAV-SfM-estimated grass height, temperature, precipitation, and vegetation indices as predictor variables. The results indicated that, even under optimal conditions, the achieved accuracy was insufficient for direct application in breeding and productivity testing trials, highlighting the limitations of UAV-SfM for yield estimation at present. For individual selection, Dactylis glomerata was assessed by comparing individual plant heights derived from UAV-SfM to rG, an established index known to correlate with breeder scores. A strong positive correlation (r > 0.8) was observed between rG and UAV-SfM-estimated individual plant heights, suggesting that UAV-SfM-derived height is potentially useful for individual plant selection. Our findings suggest that while UAV-SfM shows potential for individual plant selection in grass breeding, further methodological refinement is required before it can be reliably employed for yield estimation.