Classification of Fresh Vegetables Through Deep Learning and Neural Networks

Abstract

In the modern age, it is customary to employ machine learning to perform tasks characteristic of human intelligence. Machines are trained to use their own senses like planning, pattern recognition and image recognition. The main objective of the project is to use machine learning to identify freshness of vegetables and weed out rotten ones. The machine needs to be trained with huge amounts of data so that the model can infer certain relations and common features related to the objects. To perform this task, we need to import a set of libraries and split our data into the training set and the test set. A convolutional neural net comprises of three parts, namely the convolution, polling, flattening, respectively. This process is used to extract the feature set from the input image. Here, the spatial relationship between the pixels is preserved. While training your data, you need a lot of data to train upon. A large number of people are deployed for sample collection. To prevent overfitting, we use data augmentation. This is followed by training the model. With increasing number of epochs, the accuracy will increase. Now, we test a random image and get the machine to comment on its freshness.