Abstract: Disclosed is technology for edge-based assessment of conditions associated with a location. A method can include receiving, from a backend computer system by an edge computing device, compressed machine learning models and instructions to execute an image processing pipeline at the edge, executing, at the edge, the instructions to run the image processing pipeline, capturing images of products and fixtures associated with the location, and for each image: applying a product recognition model, receiving model output indicating products recognized in the image, applying, at least a portion of the compressed machine learning models to the model output, in response, determining conditions associated with the location, determining whether the conditions satisfy one or more KPI analyses or predetermined objectives for the location, and generating recommendations for improving the conditions.

Abstract: Disclosed is technology for edge-based assessment of conditions associated with a location. A method can include receiving, from a backend computer system by an edge computing device, compressed machine learning models and instructions to execute an image processing pipeline at the edge, executing, at the edge, the instructions to run the image processing pipeline, capturing images of products and fixtures associated with the location, and for each image: applying a product recognition model, receiving model output indicating products recognized in the image, applying, at least a portion of the compressed machine learning models to the model output, in response, determining conditions associated with the location, determining whether the conditions satisfy one or more KPI analyses or predetermined objectives for the location, and generating recommendations for improving the conditions.