The project proposal aims to develop a machine learning algorithm applied to object recognition. Specifically, the machine must be able to identify and classify into specific groups, in a fully automated manner, each micro-component belonging to electronic boards to be analyzed (resistors, transistors, capacitors, etc.).
The initial phase of the project will include studying the current problem and defining the most appropriate solution. This will be followed by structuring the database and writing the code that will allow the training and testing phases to be carried out. These phases consist of: 1) submitting the database useful for learning the distinctive characteristics of the population to be analyzed to the automatic routine, and 2) testing the effectiveness of the learning by operating on new data, evaluating the result, making any necessary modifications, and performing the so-called fine-tuning, i.e., the precise determination of control parameters.
The proposal will conclude when the algorithm is implemented and adequately optimized for regular use by the end user.
Thanks to this technology, STC will be able to achieve a dual advantage during its cost engineering activities: reduction of the time required for the classification of electronic components, which is currently done manually, and greater precision in identification.
