WHAT WAS USED:

The Keras framework was used in Python.

ALGORITHM:
The CRNN model, or Convolutional Recurrent Neural Network, uses a CNN (convolutional neural network) to extract visual features, which are reshaped and fed to an LSTM (long short term memory network). The output of the LSTM is then mapped to the character label space with a dense layer

NETWORK:
The network was trained, optimised and tested using various open datasets provided by Kaggle.

PURPOSE:
The purpose of this neural network is to “read” and “interpret” in real time the labels of products of different types and not ordered but present on the same conveyor belt and then command downstream diverters to sort them at the respective collection points.