Normalized AlexNet Deep Learning based Edibility Classification of Sporocarp

The manufacturing and dissemination of spores is the main purpose of sporocarps which is the specialized type of configuration found in freshwater plants of the Salviniales family. Though sporocarp are mainly used for reproduction, some of them are consumable by the animals and birds for the extraction of rich nutrients. Sporocarp categorization is an extremely difficult task as various species of sporocarp share significant appearance. Some sporocarp may be poisonous that lead to brain damage and infection to the animals. This paper proposes Normalized AlexNet that classifies the edible and poisonous sporocarps with high efficiency. For the sake of this investigation, 1800 sporocarp images from the Sporocarp edible Dataset were used. The best CNN model was chosen by applying the Sporocarp edible dataset to the current CNN models. With an accuracy of 89.67%, AlexNet performs better than the other CNN models. AlexNet has been chosen to fine-tune its single dense layer at this level. The AlexNet typically consists of two blocks with a single convolution including a max pooling layer, one block with three convolutions along with a max pooling layer, and one last block with three dense layers. The proposed Normalized AlexNet was now suggested by modifying the second dense layer with a single block of batch normalization, relu optimization, and drop out with a vector size of 4096. The suggested Normalized AlexNet and the current CNN have both been fitted with the Sporocarp Edible Dataset. The demonstration illustrates that the Normalized AlexNet model performs better in the edibility categorization of sporocarp, with a high accuracy of 98.76%.