FINE GRAINED CLASSIFICATION OF MONKEY SPECIES

Abstract

The aim of this project is to create an image classification algorithm that will aid in the classification of monkey species, which will be useful in the monitoring of species that are on the verge of extinction, as well as identifying places where there may be a risk of human contact with such species. The pre-processing stage, segmentation, and feature extraction are all stages of image processing that will be studied and discussed. Google Collab will be used to complete the final programming process, which will be done in Python. The programme for this project was created using the Artificial Neural Network technique. The key benefit of using this method is that it allows for the extraction and identification of features that are appropriate for character recognition. Different neural network models are discussed, with the Error-back propagation algorithm being used because of its ability to shape internal representations of features in classification. After a series oftrials and errors, a suitable set oftraining parameters is determined, and a network structure with 69 input neurons, 324 neurons for both hidden layers, and 38 neurons for the output layer with 69 input neurons is formed. The system begins by performing a pre-processing of the captured image, which includes thresholding, inverting, and smoothing. The method also includes filtering, segmentation, resizing, and feature extraction. The feed forward method is then used to generate an output matrix via the network. The recognised character can be calculated using the output matrix. This framework is intended to personalise the network for each user. The report also includes recommendations for future growth and conclusions