As part of their efforts to promote agtech, Western Growers is developing a freely available image library of key specialty crops. The idea is to create a database of labeled/annotated images that startups and researchers can use as training sets to develop AI software for automated/robotic machines. Sets of images of mature iceberg, romaine, broccoli, cauliflower and strawberry crops have been completed and are available at https://github.com/AxisAg/GHAIDatasets/tree/main/datasets. These images are useful for creating AI models for automated harvesting machines.

Last season, in collaboration with Axis Ag, Inc., we worked to expand the database to include images of crops at all growth stages. This will allow users to develop AI tools for crop thinning, weeding and crop health monitoring. Francisco Calixtro, a UofA Yuma student majoring in Ag Systems Management, spent the winter collecting images of various vegetable crops throughout their growth cycle using an Amiga¹ (farm-ng, Watsonville, CA) robot equipped with a camera. These images will be sorted, labeled and uploaded to the image library. We’ll announce when these data sets become available. 

Special thank you to Jason Mellow, Axis Ag, Inc. and to the many growers who allowed us to capture images of their fields.

Fig. 1. Francisco Calixtro, UofA Yuma student, operates an Amiga¹ (farm-ng,
Watsonville, CA) robot equipped with a camera to capture images of various
vegetable crops at different growth stages. Images will be labeled and
uploaded to a freely available image library to facilitate development of AI
software for automated/robotic machines. (Photo credits: Jason Mellow and
Francisco Calixtro)

[1] Reference to a product or company is for specific information only and does not endorse or recommend that product or company to the exclusion of others that may be suitable.