Flooring design plays a critical role in broiler welfare, hygiene, and productionefficiency. To address the limitations of conventional litter and perforated flooring systems, a non-perforated grooved floor combined with robotic cleaning has been developed.However, it remains unclear whether groove width influences broiler occupancybehavior and, consequently, the engineering design of the flooring system. This studydeveloped a computer vision framework to quantify broiler spatial distribution andevaluate whether groove width influences floor preference throughout a six-weekproduction cycle. Eighty day-old male broilers (Yield Plus × Ross 708) were reared in a6 ft × 18 ft floor pen with voluntary access to 3 semicircular groove widths (one-quarter,three-eighths, and one-half inch). Broiler occupancy was quantified from overheadRGB videos using a YOLO11-based detection framework and analyzed using a dailyoccupancy rate over both the full pen and an edge-removed region to account foredge-huddling behaviors. A detector trained on images spanning across all growthstages outperformed age-specific models, indicating the advantage of incorporatingmorphological diversity throughout the production cycle. Feeder and drinker occupancymoderately predicted bird distribution within the edge-removed region (R2 = 0.592) butnot the full pen (R2 = 0.037), where perimeter congregation dominated. Broilersexhibited no significant preference among the three groove widths (p > 0.05), and thislack of preference remained consistent throughout the production cycle. These findingsindicate that groove width can be selected primarily according to engineeringconsiderations, such as higher feces holding capacity and cleaning efficiency, withoutadversely affecting broiler occupancy. The proposed computer vision frameworkprovides an automated and scalable approach for evaluating livestock housing designsand supports the development of robotic-cleanable poultry production systems.