/vedic-symbol-recognizer-with-deep-learning

A DEEP LEARNING BASED MULTIPLE IMAGE CLASSIFIER USING CNN+RELU+ADAM OPTIMIZER AND GUI.

Primary LanguageJupyter Notebook

DOMAIN OF DATASET CONTAINS SYMBOLS OF VEDIC PERIOD SUCH AS :-

SYMBOLS:-

  1. OM

  2. SWASTIKA

  3. SRI-YANTRA

WORSHIPPING TOOL:-

  1. FIVE DIFFERENT TYPES OF SANKHA

  2. ANCIENT DESIGN KALASH

  3. DIFFERENT DIYAS

In the project:-

Project is implemented using:-

  1. keras, Tensorflow, OpenCV, tkinter

Steps:-

  1. Data preprocessing (RGB to Grey & resizing)
  2. Training the model using CNN (test, train split, epoches)
  3. Getting done by plotting graph
  4. Use of 2X3 model for better accuracy
  5. GUI model

problems faced:-

  1. Overfitting
  2. problem with accuracy
  3. settling the dataset

further work:-

  1. deployment & enhancement of domain of dataset.