To investigate the infl uence mechanism of plant landscape color design on public visual perception, ths study employed a multidisciplinary approach integrating eye-tracking, Principal Component Analysis (PCA), and the HSB color model. Focusing onplant landscapes within Shanghai universities, visual data was collected using Tobii eye-tracking equipment. Color attributeswere quantified utilizing the HSB model, and PCA was employed to examine the correlations between color factors and eyemovement metrics. The results demonstrated that: (1) A composition of 709% evergreen plants and 30% seasonal floweringplants signifcantly reduces repeated fxations, enhances overall harmony, and improves perceptual eficiency and pleasure. (2Strategically introducing accent colors, such as red fowers, within a green background effectively guides visual fow, reducesvisual stagnation, and increases the landscape's emotional valence. (3) Distributing high-saturation plants across distinct visuafocal points avoids visual oppression and enhances spatial exploration interest, (4) Significant gender difierences exist in visuaperception, females exhibit a “'scanning-style search" pattem, while males prefer a “focus-style search" pattern. Based on thesefndings, a data-dhiven hierarchical design strategy for plant landscape color structure is proposed to provide scientifc guidancefor optimizing urban plant landscape design.