Sustainable cultivated-land governance requires spatial information that is accurate, traceable, and suitable for expert review. This study developed a human-in-the-loop GeoAI–GIS workflow that transforms high-resolution semantic segmentation outputs into image-derived land-cover objects, transparent review priorities, and structured spatial evidence records. The Cultivated Land Information Management Transformer (CLIM-Former), trained on LoveDA and transferred to Gaofen-2 imagery of Mancheng District without target-area fine-tuning, provided the semantic basis for downstream object construction. Connected-component analysis generated measurable objects with semantic, geometric, locational, contextual, and provenance attributes, while interpretable rules prioritized objects for review and deterministic templates produced standardized records and verification case sheets. Across three training seeds, the guard-enabled downstream configuration achieved a mean mIoU of 56.27 ± 0.10% and an agriculture IoU of 63.12 ± 0.90%. In a reproducible 100-tile subset, 3352 objects were retained, and 1705 were prioritized. Expert acceptance was 83.00% for 100 sampled prioritized objects and 37.00% for 100 non-prioritized controls, corresponding to a 2.24-fold enrichment with a 95% confidence interval of 1.71–2.94 and a significant between-group difference according to Fisher’s exact test at p < 0.001. The generated records achieved 100% key information coverage, 95.05% factual consistency, and a mean suitability score of 4.28/5. The workflow provides a transparent and reviewable link between remote-sensing interpretation and cultivated-land information management while preserving expert authority over management decisions.

High-resolution remote sensing semantic segmentation requires the joint modeling of local details, global semantics, and height-derived geometric structures, and it provides an important basis for urban object mapping, land-cover analysis, and fine-grained spatial understanding. However, in complex urban scenes, fine-grained boundaries, small objects, inter-class similarity, and spectral confusion can still weaken the stability of pixel-level prediction. To enhance discriminative dense feature representations in high-resolution remote sensing images, we propose HDSMNet, a dual-branch multimodal semantic segmentation network designed for optical–nDSM data. The network separately extracts appearance and semantic features from optical imagery and height–structural features from nDSM, and introduces a Height-Guided Sparse Cross-Modal Fusion (HGSCF) module. Rather than treating nDSM as an additional feature source for generic fusion, HGSCF derives contextual representations, local feature contrasts, and structural-discontinuity cues from encoded nDSM features and uses them to guide sparse anchor-based interaction between optical and height features. This design enhances discriminative dense feature representations through interaction with a compact set of geometry-guided anchors. To complement HGSCF at the output stage, HDSMNet further adapts a Context-Guided Refinement (CGR) path that combines intermediate-response-guided contextual aggregation with dynamic feature modulation. This supplementary path recalibrates decoder features for output refinement. Experiments on the ISPRS Potsdam and Vaihingen datasets show that HDSMNet achieves mIoU values of 86.57% and 84.22%, respectively; ablation results further identify HGSCF as the main contributor to the observed improvement.