Walking on uneven surfaces alters foot joint kinematics and challenges gait stability. The intricate joint coupling relationship of the human foot, which is essential for biomechanical adaptations, particularly when encountering uneven surfaces, remains unclear. This study focused on quantifying foot joint coordination on cross-slopes using a markerless gait analysis method. Twelve healthy subjects performed walking tests on a gait platform under level, 8° everted and 8° inverted surface conditions. Segmental motion between rearfoot, midfoot, forefoot, and hallux were analyzed using a point cloud-based multi-segment foot model (MSFM). Adaptive changes of multi-segmental foot kinematics and inter-joint coupling relationships were compared across different cross-slope conditions. The results indicated that the rearfoot dominated frontal plane motion during everted surface walking in both middle and late stance, while the forefoot and midfoot dominated during inverted surface walking, respectively. In contrast to level walking, the sagittal-plane motion of the midtarsal joints during everted and inverted surface walking did not significantly contribute to foot kinematics at push-off. Further analysis reveals that significant variabilities exist in joint coupling behavior at different phases of the cross-slope walking.These findings demonstrate the effectiveness of the proposed method in detecting the complex inter-joint kinematics and coupling patterns of the human foot during cross-slope walking. The results provide insights into the kinematic changes of foot joints for terrain adaptation in uneven walking surfaces and advocate the application of novel motion analysis methods for tracking natural gait patterns.