This is a multi class image classification model trained using convoltuional neural networks on a custom images dataset containing images of 13 people( classes).
https://github.com/agarg2004/Image-classification-using-cnn/blob/main/cnn_updated_code The above file contains the final code of convolutional neural network implementation on the custom images dataset. It can be run on any python interpreter containing the required dependencies.
The images_dataset_mlsc contains the unlabelle, unanotated images dataset. Annoated images contains only those images which have been labelled and the labels have been stored in dataset_new.csv. The reference_images and test_images folder can be used for testing.
This model is having an accuracy of 70% which can be increased by hyperparameter tuning like modifying the learning rate, batch size, number of filters and performing preprocessing of data like rehaping, resizing, grayscaling, etc.
