Desk assessment activities for digital systems and/or data
Summary
Desk assessment activities for digital systems and/or data
S00179
Version: 1.10
- arable farming
- horticulture
- viticulture
- desk assessment
- design/documentation; other (including: nothing)
- Location: remote
- Offered by: POLIMI; UNIMI
Description
This service is offered as a first step towards testing features of a technical system or dataset provided by the customer. Examples of digital systems to be tested in this context may include, for instance, AI models for detecting grape bunches in vineyards, autonomous navigation algorithms to effectively traverse plant rows, as well as adaptive yield and ripeness estimation models.
Given a set of features of the digital system that the customer is interested in testing (e.g., a model’s ability to accurately discriminate weeds from crops given a set of images), we will perform a detailed mapping of the customer’s existing system and data to their test requirements and desired Technology Readiness Level (TRL).
The objective of this analysis is to highlight any features of the provided system (i.e., functionalities of the AI model and/or characteristics of the dataset) that will need to be further enriched or fully implemented, to help the customer’s solution reach their intended TRL and therefore enable performance testing to gauge its ability to attain the customer’s goals. The subsequent testing activity thus enabled can be performed by the customer on their own or, if required, via AgrifoodTEF Services, such as S00176, S00177 and S00178.
Example service: The customer is interested in testing the performance of a Computer Vision model to segment and count individual grape bunches in vineyards. Preliminary tests at the customer’s premises indicate that the current solution is not sufficiently robust when bunches are occluded by foliage, while the customer’s goal is to test and characterise the system’s performance in general scenarios and use the results to generate interest in potential customers and investors. To reach the point where the system is testable in a meaningful way, it is therefore necessary for the customer to further develop it. Through the desk assessment, we identify key elements of the existing Computer Vision model and implementation where the user can apply further refinement to maximum effect: e.g., to help mitigate segmentation errors. Desk assessment can also suggest alternatives among existing models and approaches that may be more suitable for the customer’s use case (e.g., including a 3D reconstruction module from multiple viewpoints wherever grapes are only partially visible). All the elements of the analysis provided by Service S00179 will be accompanied with quantitative figures to help estimate the return of investment for the customer: e.g., acquisition of the additional data recommended to bring the solution up to the specified performance level, if performed by AgrifoodTEF via Service S00113, will require 3 person months.