In the field of Deep Learning and Computer Vision, there have been several advancements. These models helped to produce state-of-the-art results on tasks like Image recognition and Image Classification, especially with the advent of extremely deep Convolutional Neural Networks. As a result, Deep Learning Architectures have evolved through time (adding more layers) to tackle increasingly complicated problems, which has aided in boosting Classification and Recognition task performance as well as making them more resilient. Deep Learning models are used to train the model from scratch or also use the existing pre-trained models using Transfer Learning method. Our paper proposes the comparative analysis of pre-trained models using Transfer Learning method for GoogLeNet, ResNet50, ResNet101 and VGGNet models. The models are trained on user defined dataset to recognize hand written digits in Telugu Language.

Computer Vision ; Convolution Neural Network ; Hand Written Character; Image Recognition ; Transfer Learning