About this Research Topic
Small-scale (or small-holder) farms support the livelihoods and food security of people in many regions, but are often under-productive and typically exposed to substantial risk from climatic and other natural hazards. The socio-economic importance of these systems is likely to remain high, and even grow, as the number of small-scale farms increases as part of broader agriculture expansions underway in much of the developing world. Given their importance, there is a pressing need for solutions to boost both the productivity and resilience of small-scale agricultural systems. Key to addressing this need is the development of accurate, reliable, and scalable methods for measuring their productivity, distribution, and management, and how these might be changing over time. Developing such methods is a particular challenge for small-scale farming systems, as they are typically found in regions where background information is sparse, and their characteristics are notoriously difficult to estimate with remote sensing.
Advances in Artificial Intelligence (AI) and other technologies are dramatically increasing the ability to monitor and measure small-scale farms. These advances are being driven by rapid improvements in machine learning techniques, particularly deep learning, as well as the increasing capability and falling costs of collecting data at spatial and temporal scales relevant to small-scale farming over large extents. These include imagery collected from small satellites and other new Earth Observation constellations, as well as in situ sensors and in field observations enabled by smartphone technology, amongst many others.
In this Research Topic, we invite submissions that demonstrate how AI can be combined with new data collection technologies to improve the ability to monitor, measure, and/or manage small-scale farms. We are particularly interested in approaches that have only become feasible in the past 3-5 years. Some potential topics of interest include:
• Large area mapping of small-scale croplands using satellites and deep learning models
• Application of AI to satellite time series to map land use and land cover change in small-scale systems, including crop types and field dynamics
• AI-assisted upscaling of on-farm sensor observations through drones to satellites
• Multi-sensor data fusion approaches to enhance AI-based agricultural mapping and monitoring
• AI-based combination of multi-source, variable quality field observations to detect agricultural signals (e.g. pest events, management activities)
• Combining AI and process-based models to improve estimation of smallholder yields
Papers may cover a variety of geographies and farming systems, as well as a broad range of scales and observational perspectives. We commit to reviewing each submission in a thorough, fair, and constructive manner. We strongly encourage authors to publish their code and data along with their accepted papers. We particularly encourage the submission of training datasets to the Radiant MLHub repository.
Upon finalization of this collection, we will select one of the published submissions for the best paper award, which in addition to the recognition will include a waiver for a future submission in Frontiers in Artificial Intelligence.
Frontiers is pro-actively committed to financially support its community on a global scale. Therefore, we encourage authors lacking support for Open Access publishing to request a fee support by filling out this application form, upon submission. Fee support allocation is based on criteria that include author publication history, country, and previous support allocated. Submitting authors should also check Frontiers’ Institutional Agreement list to check whether their Institution has an existing agreement with Frontiers. Authors affiliated with such universities or institutes will be released of some/all responsibility for Article Processing Charges (APCs) - depending on the agreement with their institution.
Participants in the Agriculture Vision workshop held in conjunction with the IEEE CVPR 2022 conference are encouraged to submit extended versions of short conference papers to this topic.
The Topic Editors would like to acknowledge Dr. Catherine Nakalembe from the University of Maryland for providing the cover image.
Advances in Artificial Intelligence (AI) and other technologies are dramatically increasing the ability to monitor and measure small-scale farms. These advances are being driven by rapid improvements in machine learning techniques, particularly deep learning, as well as the increasing capability and falling costs of collecting data at spatial and temporal scales relevant to small-scale farming over large extents. These include imagery collected from small satellites and other new Earth Observation constellations, as well as in situ sensors and in field observations enabled by smartphone technology, amongst many others.
In this Research Topic, we invite submissions that demonstrate how AI can be combined with new data collection technologies to improve the ability to monitor, measure, and/or manage small-scale farms. We are particularly interested in approaches that have only become feasible in the past 3-5 years. Some potential topics of interest include:
• Large area mapping of small-scale croplands using satellites and deep learning models
• Application of AI to satellite time series to map land use and land cover change in small-scale systems, including crop types and field dynamics
• AI-assisted upscaling of on-farm sensor observations through drones to satellites
• Multi-sensor data fusion approaches to enhance AI-based agricultural mapping and monitoring
• AI-based combination of multi-source, variable quality field observations to detect agricultural signals (e.g. pest events, management activities)
• Combining AI and process-based models to improve estimation of smallholder yields
Papers may cover a variety of geographies and farming systems, as well as a broad range of scales and observational perspectives. We commit to reviewing each submission in a thorough, fair, and constructive manner. We strongly encourage authors to publish their code and data along with their accepted papers. We particularly encourage the submission of training datasets to the Radiant MLHub repository.
Upon finalization of this collection, we will select one of the published submissions for the best paper award, which in addition to the recognition will include a waiver for a future submission in Frontiers in Artificial Intelligence.
Frontiers is pro-actively committed to financially support its community on a global scale. Therefore, we encourage authors lacking support for Open Access publishing to request a fee support by filling out this application form, upon submission. Fee support allocation is based on criteria that include author publication history, country, and previous support allocated. Submitting authors should also check Frontiers’ Institutional Agreement list to check whether their Institution has an existing agreement with Frontiers. Authors affiliated with such universities or institutes will be released of some/all responsibility for Article Processing Charges (APCs) - depending on the agreement with their institution.
Participants in the Agriculture Vision workshop held in conjunction with the IEEE CVPR 2022 conference are encouraged to submit extended versions of short conference papers to this topic.
The Topic Editors would like to acknowledge Dr. Catherine Nakalembe from the University of Maryland for providing the cover image.
Keywords: small-scale agricultural systems, small-scale farming systems, Artificial Intelligence, remote sensing, deep learning, small satellites, field observations, Large area mapping, small-scale croplands, satellite time series, land use and land cover, Multi-sensor data, agricultural mapping and monitoring
Important Note: All contributions to this Research Topic must be within the scope of the section and journal to which they are submitted, as defined in their mission statements. Frontiers reserves the right to guide an out-of-scope manuscript to a more suitable section or journal at any stage of peer review.
