Accepted_test
The aim of this work is to develop an accurate and efficient method for automatic detection and counting of wheat spikelets in images. In this research, several approaches to solve the problem have been considered: (1) using a neural network with YOLO architecture, with bounding squares with fixed side lengths that had spikelet centers at their center; (2) using a neural network with U-Net architecture trained on binary masks using binary cross-entropy; and (3) a neural network with U-Net architecture trained on masks from a normal bivariate distribution using Kullback-Leibler divergence as the loss function. The best results were demonstrated by the model with U-Net architecture trained on normal masks using Kulbak-Leibler divergence as a loss function.