/Multi-Class-Image-Classification

Multi-class image classification on natural images using Neural Network

Primary LanguageJupyter Notebook

Multi-Class-Image-Classification

Natural Image Classification (Kaggle Challenge)

Here is the task-

  1. Downloading the dataset from https://www.kaggle.com/prasunroy/natural-images
    This dataset contains 6,899 images from 8 distinct classes compiled from various sources (see Acknowledgements). The classes include airplane, car, cat, dog, flower, fruit, motorbike and person.
  2. Designing a suitable neural-network model to classify these images.
  3. Randomly selecting 20% of the images as train set, training the model with the rest 80% images.
  4. Reporting classification accuracy.
  5. Showing some images that are correctly predicted. Also, showing some images that are not correctly predicted.

Python Dependencies:

NumPy
Pandas
Sklearn
Tensorflow
Matplotlib
Keras

Was inspired from-
https://www.kaggle.com/wl0000000e/get-89-val-acc-within-20-epoches

--Thanks a lot for viewing

N.B.: Jupyter-notebook files are large and may not be displayed properly in GitHub Use the online Notebook viewer: https://nbviewer.jupyter.org/ to view the notebooks.