Dogra script which was once used to write Dogri language in the state of Jammu and Kashmir and parts of adjoining states has failed to see any remarkable work towards Dogra handwritten text recognition though reasonable work has been done for text recognition of various other scripts of Indian subcontinent and scripts at the world level. Handwritten Text Recognition is a demanding area of research since many years as the information available in handwritten form is prone to loss, mutilation and also could not be used to its full potential owing to non-understanding by the machine. Absence of standardised Dogra script dataset was one of reason for negligible work in Dogra handwritten text recognition. After creating a Dogra Character Dataset of 39000 characters, various existing classification algorithms/approaches such as Support Vector Machine (SVM), Random Forest, Convolution Neural Network (CNN), K Nearest Neighbours (kNN) and Pre-trained VGG16 CNN were deployed and recognition accuracy analysed. The activation and optimized parameters were affected to further enhance the performance. The testing parameter’s result viz accuracy, precision and recall reveals that pre-trained and fine-tuned VGG16 with RMSprop optimizer out performs other algorithms and gave 97% accuracy with reduced time consumption over and above other approaches.