Gourd Vegetable Detection by Deep Learning Approach

Abstract

The local market in Bangladesh offers a diverse range of around 100 varieties of vegetables for purchase and sale. The nomenclature of vegetables exhibits regional variations, leading to difficulties in identification for individuals. Amongst these, the gourd vegetables belonging to the Cucurbitaceous family, namely Sponge Gourd, Ridge Gourd, and Snake Gourd, Bottle Gourd, Bitter Gourd pose the greatest challenge due to their strikingly similar structural characteristics despite significant differences in quality. This study focuses on the application of image processing techniques for the identification of three specific vegetables. The utilization of image processing techniques in the field of agriculture is experiencing a steady growth in recent times. The utilization of Quality Test in image processing facilitates the diagnostic procedure and serves as a means to discern several categories of vegetables by analyzing their size, shape, and color. The suggested methodology starts with the acquisition of picture data. A total of 10000 photos of Sponge Gourd, Ridge Gourd, and Snake Gourd, Bottle Gourd, Bitter Gourd were obtained in the agricultural field. Upon the completion of image processing, I conducted training and testing utilizing the model 3 architectures. The obtained accuracy rate was 99% in both cases.