Cultivated land is crucial for food production and security. In complex environments like mountainous regions, the fragmented nature of the cultivated land complicates rapid and accurate information acquisition. Deep learning has become essential for extracting cultivated land but faces challenges such as edge detail loss and limited adaptability. This study introduces a novel approach that combines geographical zonal stratification with the temporal characteristics of medium-resolution remote sensing images for identifying cultivated land. The methodology involves geographically zoning and stratifying the study area, and then integrating semantic segmentation and edge detection to analyze remote sensing images and generate initial extraction results. These results are refined through post-processing with medium-resolution imagery classification to produce a detailed map of the cultivated land distribution. The method achieved an overall extraction accuracy of 95.07% in Tongnan District, with specific accuracies of 92.49% for flat cultivated land, 96.18% for terraced cultivated land, 93.80% for sloping cultivated land, and 78.83% for forest intercrop land. The results indicate that, compared to traditional methods, this approach is faster and more accurate, reducing both false positives and omissions. This paper presents a new methodological framework for large-scale cropland mapping in complex scenarios, offering valuable insights for subsequent cropland extraction in challenging environments.

As a commercially and medicinally significant species extensively cultivated across Asia, Zanthoxylum bungeanum Maxim. (ZBM) demands precise spatial mapping to underpin regional economic development and ecological management. Although satellite image time series (SITS) coupled with machine learning have advanced crop-specific identification, accurately delineating ZBM in mountainous smallholder landscapes remains problematic. Similar agronomic practices and microclimatic variability intensify spectral confusion between ZBM and its co-cultivated counterparts, while shifting agricultural policies and diverse crop rotations complicate the establishment of fixed classification schemes and render manual annotation prohibitively laborious. To address these challenges, this study proposes the Phenology-Stage Aware Temporo-Spatial Fusion Transformer with Deep One-Class Loss (PTSFOC), a novel framework for high-precision ZBM mapping using SITS data. PTSFOC addresses spectral ambiguity through the Multi Phenology-Stage Feature Extraction (MPSE) branches and Cross-Phenology-Stage Fusion (CPSF) module, which jointly model spatiotemporal signatures across ZBM's key growth phases (e.g., fruit expansion and ripening) to enhance discriminative capacity. To alleviate labeling burdens, a deep one-class risk estimator trains the network using only positive and unlabeled samples, obviating the need for negative class definitions. Applied to Jiangjin District in Chongqing Municipality, a typical mountainous smallholder region, PTSFOC achieves an F1-score of 0.952, outperforming existing one-class classification methods. This work not only delivers a robust tool for mapping ZBM in complex agricultural landscapes but also offers a transferable strategy for other specialty crops plagued by spectral ambiguity, thereby informing targeted agricultural and ecological policy.