IDENTIFICATION OF EDIBLE WILD PLANTS THROUGH IMAGE ANALYSISUSING SYSTEM AND NETWORK SOFTWARE

Abstract

The paper analyzes the current state of the development of technologies and technical tools, with a focus on the widespread adoption of digital solutions utilizing neural network tools as effective instruments in the context of the Fourth Industrial Revolution. The feasibility of applying digital solutions in the food industry is highlighted, particularly at the stages of raw material supply as a product of agricultural production. Common types of agricultural food products include wild plants, for which strict requirements are imposed during harvesting and sorting due to the presence of dangerous look-alikes.

A method for identifying edible wild plants based on image analysis is presented. The method is based on neural network tools and includes input data processing, a recognition step, and result generation. The input data consists of an image of a wild plant. The trained model contains all necessary parameters determined during training on a relevant dataset. Procedurally, the identification process involves four steps. In the first step, the model is initialized, and the availability and correctness of loading the neural network model are verified. The second step involves image processing, including loading, scaling, normalization, and converting it to the appropriate formats. In the third step, classification is performed and the result is generated. In the fourth step, the identified wild plant is classified based on the obtained probability, and a decision is made regarding the specific wild plant recognized.

To ensure the practical value of the developed method, a corresponding software solution was created. Its efficiency is determined by the degree to which system software capabilities are utilized, while its integration into modern general-purpose business systems depends on the proper use of network software.

The results of testing the developed method and its software implementation are presented using one of the most common edible wild plants as an example.