Naruto Hand Sign Prediction using CNN

A custom CNN to classify Naruto Hand Signs. The data set contains roughly around 2100 images and I used a test split of 0.2 and validation split of 0.2 in the train set. Achieved validation accuracy of 0.95 and test accuracy of 0.96

Dataset: Kaggle Link

Model Summary

Plots

Download model from here

Instructions to Load the Model

To load model type the following code

import pickle

#filename - path to the model
loaded_model = pickle.load(open(filename, 'rb'))

Note

The html design of the current app is taken from this repo.