Abstract:Accurate crop mapping provides important information for government decision-making and agricultural management. Recent advancements in deep learning have significantly enhanced the capabilities of remote sensing-based crop mapping. In this study, we developed a new deep learning approach, DSH, which comprises three modules, DeepLabV3+ (D), channel self-attention (S), and histogram matching (H). The DeepLabV3+ module learns the spatial distribution of crops in the spatial and spectral dimensions. The channel self-attention module was used to enhance the weights of the important features. The histogram-matching module addresses the domain gap between the training and testing images and enhances the transferability of the DSH. In addition, the input data of the DSH are monthly synthesized Sentinel-2 data, thus relaxing the data collection requirements. The performance of the DSH was evaluated and benchmarked against HRNetV2, DeepLabV3+, and random forest (RF) at eight sites in the U.S. China, and France with different farming structures, climate conditions, and landscape complexities. Temporal transfer revealed that the classification performance of DSH improved as the growth season progressed, peaked in the peak growth season (August), and then declined at each study site. The DSH model trained during the peak growth season demonstrated superior temporal transferability. Spatial transfer experiments demonstrated that transferring DSH to sites with similar planting structures achieved higher precision than full spatial transfers. Spatiotemporal transfer experiments conducted at eight sites over three years (2020–2022) demonstrated an average overall accuracy (OA) of 88.1% for DSH, highlighting the importance of similar planting structures for effective spatiotemporal transfer. DSH outperformed HRNetV2, DeepLabV3+, and RF in OA, producer's accuracy, and user's accuracy at all eight study sites. Notably, DSH exhibited comparable performance during both the peak and full growth seasons, with an average OA of 93.9%. The classification accuracy of test images after histogram matching remained stable from 2020 to 2022, whereas the accuracy of raw test images fluctuated annually. This indicates that histogram matching is effective in improving the transferability of DSH. Embedding the channel self-attention module into the position of the high-level features had a more significant impact on improving DSH's performance than other configurations. Visualizing features from the channel self-attention and high-level layers revealed that integrating the channel self-attention module with DeepLabV3+'s high-level features significantly improved crop mapping by enhancing relevant semantic information. Overall, DSH exhibits robust spatiotemporal generalizability and requires minimal input data, making it suitable for large-scale crop mapping.

CROPUP: Historical Products Are All You Need? an End-to-end Cross-Year Crop Map Updating Framework Without the Need for in Situ Samples

Deep Learning-Based Crop Mapping in the Cloudy Season Using One-Shot Hyperspectral Satellite Imagery.

Deep One-Class Crop Extraction Framework for Multi-Modal Remote Sensing Imagery.

Cross-Year Reuse of Historical Samples for Crop Mapping Based on Environmental Similarity

Early- and in-season crop type mapping without current-year ground truth: Generating labels from historical information via a topology-based approach

Tracking spatiotemporal dynamics of crop-specific areas through machine learning and statistics disaggregating

Crop Mapping Using the Historical Crop Data Layer and Deep Neural Networks: A Case Study in Jilin Province, China

Weakly Supervised Framework Considering Multi-temporal Information for Large-scale Cropland Mapping with Satellite Imagery

Croplayer-China: A 2-Meter Resolution Cropland Map of China Based on Active Learning of Segmentation with Mapbox and Google Satellite Imagery

A Novel Knowledge-Driven Automated Solution for High-Resolution Cropland Extraction by Cross-Scale Sample Transfer.

Mapping corn dynamics using limited but representative samples with adaptive strategies

DOCC: Deep One-Class Crop Classification Via Positive and Unlabeled Learning for Multi-Modal Satellite Imagery

A new transferable deep learning approach for crop mapping

A Multi-Level Fine-Grained Crop Classification Method Based on Multi-Expert Knowledge Distill

Multi-year mapping of cropping systems in regions with smallholder farms from Sentinel-2 images in Google Earth engine

DeepCropMapping: A multi-temporal deep learning approach with improved spatial generalizability for dynamic corn and soybean mapping

Crop classification methods and influencing factors of reusing historical samples based on 2D-CNN

Machine-learned prediction of annual crop planting in the U.S. Corn Belt based on historical crop planting maps

A Sampling Workflow Based on Unsupervised Clusters and Multi-Temporal Sample Interpretation (UCMT) for Cropland Mapping

Cross Domain Early Crop Mapping using CropSTGAN

Efficient Corn and Soybean Mapping with Temporal Extendability: A Multi-Year Experiment Using Landsat Imagery