innovator-arjun/multiclass-image-prediction-with-sparse-dataset

A CNN architecture was built using keras with convolution & pooling layers, Dropouts, and with a Softmax layer the last. Ensemble approach to make the model more robust and with generalization capability to the unseen data. Bayesian optimization is used for hyper parameter tuning the deep learning model.

Multiclass Image Prediction with Sparse Dataset

A CNN architecture was built using keras with convolution & pooling layers, Dropouts, and with a Softmax layer the last. Ensemble approach to make the model more robust and with generalization capability to the unseen data. Bayesian optimization is used for hyper parameter tuning the deep learning model.

In this project, we will classify from 6 classes: [ant, spider, flower, dolphin, lobster, bulldozer]. These image were hand drawn by people around the world, as part of the project Quickdraw.