Skip to content

Latest commit

 

History

History
26 lines (14 loc) · 988 Bytes

README.md

File metadata and controls

26 lines (14 loc) · 988 Bytes
Raw

MNIST-Handwritten-Digit-Clssification-using-Deep-Learning.

About the MNIST dataset:

---> It consists of 60,000 images with 28X28 dimensions and 10,000 test image data. It is a type of grayscale image and the image processing and cross-validation part is already done in this dataset.

We can find the dataset directly in the Keras library where we can use the API .https://keras.io/api/datasets/mnist/

Operations carried out:

  1. Importing the necessary dependencies (libraries).
  2. Load the dataset in the Google Colab directory.
  3. Image data analysis.
  4. Image label analysis.
  5. Building the neural network using tensorflow and keras library.
  6. Model Evaluation to check accuracy and loss on test data and also if there is any overfitting or not.
  7. Use the Confusion matrix and visualize the data using the heatmap
  8. Build a predictive system, which will take different handwritten digit images as a path and can recognize the label of the image data.