Seed Quality Prediction using Computer Vision and Convolutional Neural Networks

Knowing the quality of the seeds is the most important thing for a farmer. The quality of the seed is critical for obtaining a good yield. Farmers typically purchase seeds from companies. They will ensure that the seeds are of high quality. However, because they use manual force to collect good quality seeds, there is a chance that they will replace good quality seeds with damaged ones. The precaution to be taken is to predict the seed quality. Our goal is to provide a solution that eliminates the need for manual seed quality checks in commercial farming. Without human intervention, the detection of pure and damaged seeds requires the use of computer vision and deep learning techniques. The proposed model employs OpenCV to detect every seed grain in the seed lot, as well as a convolutional neural network to predict the quality of the detected seed grain. The model's output is a prediction of the seed lot's purity percentage. This paper suggests a solution that reduces the amount of manual labor and time required to filter out damaged seeds.