Animal-Pictures-Classification-using-CNN

CNN to classify animal pictures from the Animal-10 dataset

Topic

This project is about testing a CNN on the Animal-10 dataset, which contains 28K medium quality image of ten animals : squirrel, hen, horse, butterfly, dog, cat, cow, spider, sheep and elephant. The pictures have been all downloaded from google so they have different sizes. I will test a multilayered CNN to try and categorize the pictures, after having processed the images and done the essential transformations. And after the training phase I will go on to test my model of a test set it has never seen before. I'm hoping for an accuracy of about 80%.

Objectives

• Preprocess the data to make it machine learning ready
• Build a CNN with enough layers to identify the differences between the images
• Train and test the model and get the accuracy

Summary

• Importing libraries
• The dataset
• Image processing
• Some visualization
• Building the classifier
• Training the model
• Testing the model
• Conclusion

Libraries

• Pandas
• Numpy
• Torchvision
• Matplotlib
• Glob
• CV2

Data source

https://www.kaggle.com/datasets/alessiocorrado99/animals10?resource=download&fbclid=IwAR1Zn8vaqfGdRCEM-cRGyO46zahJ4k0g6aFcRh7oOeHKK7TL0gRzC6-n4xc