An official website of the United States government
Mapping and monitoring crops is a key step towards the sustainable intensification of agriculture and addressing global food security. A dataset like ImageNet that revolutionized computer vision applications can accelerate the development of novel crop mapping techniques. Currently, the United States Department of Agriculture (USDA) annually releases the Cropland Data Layer (CDL) which contains crop labels at 30m resolution for the entire United States of America. While CDL is state of the art and is widely used for a number of agricultural applications, it has a number of limitations (e.g., pixelated errors, labels carried over from previous years, and errors in the classification of minor crops). In this work, we create a new semantic segmentation benchmark dataset, which we call CalCROP21, for the diverse crops in the Central Valley region of California at 10m spatial resolution using a Google Earth Engine based robust image processing pipeline and a novel attention-based spatio-temporal semantic segmentation algorithm STATT. STATT uses re-sampled (interpolated) CDL labels for training but is able to generate a better prediction than CDL by leveraging spatial and temporal patterns in Sentinel2 multi-spectral image series to effectively capture phenologic differences amongst crops and uses attention to reduce the impact of clouds and other atmospheric disturbances. We also present a comprehensive evaluation to show that STATT has significantly better results when compared to the resampled CDL labels. We have released the dataset and the processing pipeline code for generating the benchmark dataset.
more »
« less
This content will become publicly available on January 23, 2026
Cyberinfrastructure for machine learning applications in agriculture: experiences, analysis, and vision
IntroductionAdvancements in machine learning (ML) algorithms that make predictions from data without being explicitly programmed and the increased computational speeds of graphics processing units (GPUs) over the last decade have led to remarkable progress in the capabilities of ML. In many fields, including agriculture, this progress has outpaced the availability of sufficiently diverse and high-quality datasets, which now serve as a limiting factor. While many agricultural use cases appear feasible with current compute resources and ML algorithms, the lack of reusable hardware and software components, referred to as cyberinfrastructure (CI), for collecting, transmitting, cleaning, labeling, and training datasets is a major hindrance toward developing solutions to address agricultural use cases. This study focuses on addressing these challenges by exploring the collection, processing, and training of ML models using a multimodal dataset and providing a vision for agriculture-focused CI to accelerate innovation in the field. MethodsData were collected during the 2023 growing season from three agricultural research locations across Ohio. The dataset includes 1 terabyte (TB) of multimodal data, comprising Unmanned Aerial System (UAS) imagery (RGB and multispectral), as well as soil and weather sensor data. The two primary crops studied were corn and soybean, which are the state's most widely cultivated crops. The data collected and processed from this study were used to train ML models to make predictions of crop growth stage, soil moisture, and final yield. ResultsThe exercise of processing this dataset resulted in four CI components that can be used to provide higher accuracy predictions in the agricultural domain. These components included (1) a UAS imagery pipeline that reduced processing time and improved image quality over standard methods, (2) a tabular data pipeline that aggregated data from multiple sources and temporal resolutions and aligned it with a common temporal resolution, (3) an approach to adapting the model architecture for a vision transformer (ViT) that incorporates agricultural domain expertise, and (4) a data visualization prototype that was used to identify outliers and improve trust in the data. DiscussionFurther work will be aimed at maturing the CI components and implementing them on high performance computing (HPC). There are open questions as to how CI components like these can best be leveraged to serve the needs of the agricultural community to accelerate the development of ML applications in agriculture.
more »
« less
- Award ID(s):
- 2112606
- PAR ID:
- 10577776
- Publisher / Repository:
- Frontiers in Artificial Intelligence, Sec. AI in Food, Agriculture and Water
- Date Published:
- Journal Name:
- Frontiers in Artificial Intelligence
- Volume:
- 7
- ISSN:
- 2624-8212
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Passive microwave remote sensing plays an essential role in providing valuable information about the Earth’s surface, particularly for agriculture, water management, forestry, and other environmental fields. One of the key requirements for precision agricultural applications is the availability of field- scale high-resolution remote sensing data products. With the recent development of reliable unmanned aircraft systems (UAS), airborne deployment of remote sensing sensors has become more widespread to provide such products. With this in mind, we developed a UAS-based dual H-pol (hori- zontal) and V-pol (vertical) polarized radiometer operating in L-band (1400-1427 MHz). The custom dual-polarized an- tenna acquires surface emission response through a software- defined radio (SDR). This SDR-based system provides full control over the data acquisition parameters such as band- width, sampling frequency, and data size. Radio frequency interference (RFI) poses a significant challenge in radiometric measurements, requiring post-processing of the full-band radiometer data to identify and eliminate RFI-contaminated measurements, thus ensuring accurate Earth emission read- ings.. In this paper, we implemented near-real-time RFI detection onboard during the flight to accelerate the post- processing. The altitude and the speed of the UAS can be varied to achieve desired ground resolution for the measure- ment. This paper presents the full custom design and develop- ment of a lightweight SDR-based UAS-borne radiometer for precision agriculture. Additionally, we introduce the concept of an agile radiometer implemented from a small UAS that can serve as a testbed for both current and future spaceborne missions.more » « less
-
Abstract ContextUnoccupied aerial systems/vehicles (UAS/UAV, a.k.a. drones) have become an increasingly popular tool for ecological research. But much of the recent research is concerned with developing mapping and detection approaches, with few studies attempting to link UAS data to ecosystem processes and function. Landscape ecologists have long used high resolution imagery and spatial analyses to address ecological questions and are therefore uniquely positioned to advance UAS research for ecological applications. ObjectivesThe review objectives are to: (1) provide background on how UAS are used in landscape ecological studies, (2) identify major advancements and research gaps, and (3) discuss ways to better facilitate the use of UAS in landscape ecology research. MethodsWe conducted a systematic review based on PRISMA guidelines using key search terms that are unique to landscape ecology research. We reviewed only papers that applied UAS data to investigate questions about ecological patterns, processes, or function. ResultsWe summarize metadata from 161 papers that fit our review criteria. We highlight and discuss major research themes and applications, sensors and data collection techniques, image processing, feature extraction and spatial analysis, image fusion and satellite scaling, and open data and software. ConclusionWe observed a diversity of UAS methods, applications, and creative spatial modeling and analysis approaches. Key aspects of UAS research in landscape ecology include modeling wildlife micro-habitats, scaling of ecosystem functions, landscape and geomorphic change detection, integrating UAS with historical aerial and satellite imagery, and novel applications of spatial statistics.more » « less
-
The successful implementation of vision-based navigation in agricultural fields hinges upon two critical components: 1) the accurate identification of key components within the scene, and 2) the identification of lanes through the detection of boundary lines that separate the crops from the traversable ground. We propose Agronav, an end-to-end vision-based autonomous navigation framework, which outputs the centerline from the input image by sequentially processing it through semantic segmentation and semantic line detection models. We also present Agroscapes, a pixel-level annotated dataset collected across six different crops, captured from varying heights and angles. This ensures that the framework trained on Agroscapes is generalizable across both ground and aerial robotic platforms. Codes, models and dataset will be publicly released.more » « less
-
BackgroundPrescribed fire is vital for fuel reduction and ecological restoration, but the effectiveness and fine-scale interactions are poorly understood. AimsWe developed methods for processing uncrewed aircraft systems (UAS) imagery into spatially explicit pyrometrics, including measurements of fuel consumption, rate of spread, and residence time to quantitatively measure three prescribed fires. MethodsWe collected infrared (IR) imagery continuously (0.2 Hz) over prescribed burns and one experimental calibration burn, capturing fire progression and combustion for multiple hours. Key resultsPyrometrics were successfully extracted from UAS-IR imagery with sufficient spatiotemporal resolution to effectively measure and differentiate between fires. UAS-IR fuel consumption correlated with weight-based measurements of 10 1-m2 experimental burn plots, validating our approach to estimating consumption with a cost-effective UAS-IR sensor (R2 = 0.99; RMSE = 0.38 kg m-2). ConclusionsOur findings demonstrate UAS-IR pyrometrics are an accurate approach to monitoring fire behaviour and effects, such as measurements of consumption. Prescribed fire is a fine-scale process; a ground sampling distance of <2.3 m2 is recommended. Additional research is needed to validate other derived measurements. ImplicationsRefined fire monitoring coupled with refined objectives will be pivotal in informing fire management of best practices, justifying the use of prescribed fire and providing quantitative feedback in an uncertain environment.more » « less
