Consultancy- Plot-Registration And Image Capture job at Alliance Bioversity & CIAT | Apply Now

The Alliance of Bioversity International delivers research-based solutions that harness agricultural biodiversity and sustainably transform food systems to improve people’s lives. Alliance solutions address the global crises of malnutrition, climate change, biodiversity loss, and environmental degradation. The Alliance works with local, national and multinational partners across Latin America and the Caribbean, Asia and Africa, and with the public and private sectors. The Alliance is part of CGIAR, a global research partnership for a food-secure future, dedicated to reducing poverty, enhancing food and nutrition security, and improving natural resources and ecosystem services.
The Alliance of Bioversity International and CIAT have initiated a project, titled Artemis. Artemis is developing AI technology to enable on-farm breeding. Crop breeding is one of the foundations of agriculture, under the current breeding systems a common assumption is that it takes 10 years to develop a new variety. Given the pace of climate change this timeframe is no longer viable. New approaches are needed to speed up the breeding cycle and enable a greater synchrony between plant breeding and on-farm conditions, shifting breeding from a majority on-station to majority on-farm process. Artemis is addressing this issue by leveraging the power of technology to enable breeding programs with a new generation of AI-powered and smartphone-deployable tools. The principal technology is computer vision phenotyping, with explorations into multimodal data that integrates speech and text to actively integrate farmers into the plant improvement process. Artemis is currently in the second investment phase, having been featured recently in Reuters.
About the Consultancy
The Plot-registration and Image Capture Consultant will be responsible for testing and building on the plot-registration methods and working with the team to test new solutions on a need-basis.

Background and Task Description
Plant breeding primarily occurs in research stations, where breeders evaluate novel plant varieties (genotypes) in small plots (approximately 4m x 3m), often ranging from several hundred to a few thousand plots.
The Artemis team has successfully developed fully executable standard operating procedures (SOPs) that capture data of sufficient quality to develop highly accurate models that measure plant performance. However, a few additional challenges are remaining around the efficiency and consistency of data capture.
Plot Registration
A variety of metadata is associated with each plot (variety type, planting practice, etc). The imagery data must be registered with this metadata for it to be useful. As the metadata is unique for each plot, data systems must be developed to efficiently capture the data. For example, standard practices are to use QR or barcode labels to register plots. However, manually scanning each QR or barcode can take several seconds or more. With only 1000 plots, ten seconds of plot registration time can total almost three hours. The current system used requires around 30 seconds of registration time. Innovative solutions must be developed to reduce this time to as little as possible.
Image Capture Methods
SOPs have been developed using manual methods (holding the phone by hand). However, this approach is labour-intensive, taking up to 5 minutes per plot, and physically demanding, making it unsuitable for efficient use at scale. To address these challenges, the team has been working on a wheeled device called 'Bruno.' Bruno consists of a simple wheeled device that can mount two phones to a central frame and be easily pushed through the fields by one person. The current Bruno can image a breeding plot in under 30 seconds.
Bruno comprises the following components:
1. A central frame connected to a small wheel and two handles.
2. A central vertical pole capable of mounting two or more phones.
3. Sidebars that can open and close to widen the angle near the wheel, facilitating observation of plant parts beneath leaves (pods and flowers).
With Bruno, the user needs to unmount and remount both phones to scan the QR code, or requires the user to manually select the plot on each phone, adding additional time to the plot registration process. In addition to this, the user needs to be able to trigger image capture consistently throughout the plot. Some image capture protocols call for imaging one side of every single plant in 2 rows, while others require taking 3-5 pictures of the whole plot from a top-down angle.
Goals
Our goal is to reduce plot registration to 10 seconds or less. Through testing with our field team, we have derived a priority list of solutions to test. These solutions will require a mix of software and hardware skills. You will need to mix and match the solutions to create a complete plot registration and image capture solution integrated across two individual phones.
• Recognition of the next plot based on wheel rotations. Given a wheel of a known diameter, each rotation corresponds to a fixed length. Can wheel rotation be used to trigger plot transition and image capture automatically? Per 2 rotations
• Leveraging existing hardware on mid-range smartphones (e.g.gyroscopes, accelerometers, GPS)
• Binary button to control yes/no or back/next in-app choices
• Small, hand-held numeric keypad to control multiple in-app choices like, yes/no, back/next, plot-selection (including typing plot-number), initiating image capture or stopping image capture, etc.