Abstract:Real-time and accurate information on fine-grained changes in crop cultivation is of great significance for crop growth monitoring, yield prediction and agricultural structure adjustment. Aiming at the problems of serious spectral confusion in visible high-resolution unmanned aerial vehicle (UAV) images of different phases, interference of large complex background and salt-and-pepper noise by existing semantic change detection (SCD) algorithms, in order to effectively extract deep image features of crops and meet the demand of agricultural practical engineering applications, this paper designs and proposes an agricultural geographic scene and parcel-scale constrained SCD framework for crops (AGSPNet). AGSPNet framework contains three parts: agricultural geographic scene (AGS) division module, parcel edge extraction module and crop SCD module. Meanwhile, we produce and introduce an UAV image SCD dataset (CSCD) dedicated to agricultural monitoring, encompassing multiple semantic variation types of crops in complex geographical scene. We conduct comparative experiments and accuracy evaluations in two test areas of this dataset, and the results show that the crop SCD results of AGSPNet consistently outperform other deep learning SCD models in terms of quantity and quality, with the evaluation metrics F1-score, kappa, OA, and mIoU obtaining improvements of 0.038, 0.021, 0.011 and 0.062, respectively, on average over the sub-optimal method. The method proposed in this paper can clearly detect the fine-grained change information of crop types in complex scenes, which can provide scientific and technical support for smart agriculture monitoring and management, food policy formulation and food security assurance.

What problem does this paper attempt to address?

The paper mainly addresses the issue of fine-grained semantic change detection in high-resolution Unmanned Aerial Vehicle (UAV) images of crops. Specifically, the research aims to solve the following key problems: 1. **Severe Spectral Confusion**: Crops at different growth stages may exhibit highly similar spectral characteristics in visible high-resolution UAV images. For example, it is difficult to distinguish whether changes have occurred in rapeseed and rice at different stages. 2. **"Salt and Pepper" Noise**: Existing Semantic Change Detection (SCD) algorithms tend to produce a large amount of noise when dealing with complex backgrounds, leading to incomplete farmland boundaries. 3. **Complex Background Features and Proportion Imbalance**: Change areas usually occupy only a part of the entire region, while the proportion of unchanged background (often containing complex distribution characteristics of non-agricultural areas) is larger. This may cause significant interference in the detection process of SCD methods. To address the above issues, the paper proposes an agricultural geographic scene-constrained farmland-scale crop fine-grained semantic change detection framework (AGSPNet). AGSPNet mainly includes three parts: - **Agricultural Geographic Scene Division Module**: This module uses multi-source open geographic data products to analyze the rules of agricultural regional differences, dividing agricultural geographic scenes with relatively consistent geographic element conditions. This helps to mitigate the impact of complex background features in large non-crop areas. - **Farmland Boundary Extraction Module**: Based on the Bi-directional Cascade Network (BDCN) and a designed boundary optimization model, this module extracts complete farmland boundaries to constrain crop semantic change results, ensuring that the final crop semantic change map matches actual needs. - **Crop SCD Module**: This module constructs a cross-attention network (Criss-Cross Attention Network, CCNet) with a pseudo-Siamese structure and integrates a change feature discrimination module to enhance the model's ability to extract crop change features from two time phases, maximizing the potential performance of the SCD network in crop mapping. In addition, the research also creates a new publicly accessible SCD dataset (CSCD) specifically for large-scale annual agricultural monitoring. This dataset is based on high-resolution UAV remote sensing images, featuring clearer dual-phase multiple crop change types and explicit "from...to..." semantic change annotation information, providing effective data support for the research of new SCD methods.

AGSPNet: A framework for parcel-scale crop fine-grained semantic change detection from UAV high-resolution imagery with agricultural geographic scene constraints

Deep Segmentation and Classification of Complex Crops Using Multi-Feature Satellite Imagery

DSCA-PSPNet: Dynamic spatial-channel attention pyramid scene parsing network for sugarcane field segmentation in satellite imagery

Spatial-Temporal Semantic Perception Network for Remote Sensing Image Semantic Change Detection

Edge-Preserving Texture-Based Semantic Segmentation for Ultrahigh Resolution Images in Agricultural Scene

A robust spectral-spatial approach to identifying heterogeneous crops using remote sensing imagery with high spectral and spatial resolutions

Smallholder Crop Area Mapped with a Semantic Segmentation Deep Learning Method.

MSCPUnet: A Multi-task Neural Network for Plot-level Crop Classification in Complex Agricultural Areas

S3ANet: Spectral-spatial-scale attention network for end-to-end precise crop classification based on UAV-borne H2 imagery

Tsanet: A Deep Learning Framework for the Delineation of Agricultural Fields Utilizing Satellite Image Time Series

Semantic Information Collaboration Network for Semantic Change Detection in Remote Sensing Images

A Generalized Approach Based on Convolutional Neural Networks for Large Area Cropland Mapping at Very High Resolution

A Novel Method for Identifying Crops in Parcels Constrained by Environmental Factors Through the Integration of a Gaofen-2 High-Resolution Remote Sensing Image and Sentinel-2 Time Series

DSHANet: dynamic sparse hierarchical attention-driven cropland change detection network with holistic complementation fusion

GLCANet: Global–Local Context Aggregation Network for Cropland Segmentation from Multi-Source Remote Sensing Images

Parcel-Level Crop Classification in Plain Fragmented Regions Based on Multi-Source Remote Sensing Images

Deep-agriNet: a lightweight attention-based encoder-decoder framework for crop identification using multispectral images

FFPNet: Fine-Grained Feature Perception Network for Semantic Change Detection on Bi-Temporal Remote Sensing Images

Using the MSFNet Model to Explore the Temporal and Spatial Evolution of Crop Planting Area and Increase Its Contribution to the Application of UAV Remote Sensing

A Study on the Object-Based High-Resolution Remote Sensing Image Classification of Crop Planting Structures in the Loess Plateau of Eastern Gansu Province

Crop Classification from Drone Imagery Based on Lightweight Semantic Segmentation Methods