Prediction of tomato seed germination from images with deep learning

This book is related to the application of deep learning in agricultural field. In fact, researches were carried out to predict the tomato seeds germinability from images with deep learning under proper growing conditions. This task was a classification task which was tricky as the classes were imbalance. To assess the models, macro f1 score was used.In the analysis, a baseline model was used which consisted of three convolutional layers and then the classification layer after a flattening operation.Some useful pre-trained models such VGG-19, ResNet-50, were also used in order to leverage the new task.Even though the classification was so difficult, the above techniques were so useful in achieving 0.52 which is better than humain guessing skills.