Personality Prediction from Handwriting using Fine-tuned Transfer Learning Models

Abstract

Experts in computational linguistics have done a number of research to identify and categorize personality characteristics at various aspects, including terms, sentences, paragraphs, and recommendations. In this research work five stages model was proposed and the experimental result was evaluated on handwriting images.This paper presents a comparative analysis of fine-tuning transfer learning convolutional neural network models such as VGG16, ResNet50 and GoogleNet for personality detection. The results of fine-tuned models are assessed using the accuracy, precision, recall and f1_score measure. From the results it was observed that ResNet50 have achieved best accuracy as compared to GoogleNet and VGG16