IAM Dataset-Based Author Identification via Convolutional Neural Networks

Optical character recognition (OCR) has emerged as a crucial technology in the field of document analysis and text processing. This paper presents a novel approach to OCR utilizing the IAM dataset, a widely recognized benchmark for handwriting recognition tasks. Our method combines deep learning techniques with advanced pre-processing methods to achieve state-of-the-art results in handwritten text recognition. The IAM dataset offers diverse and challenging handwritten samples, making it an ideal testbed for OCR systems. The proposed model is trained using a large and carefully curated subset of the IAM dataset, ensuring robustness and generalization. Furthermore, this work focuses on reducing the computational complexity and memory footprint of the OCR system, making it suitable for real-time applications. We optimize our model for efficiency while maintaining high accuracy. Our experiments demonstrate superior performance compared to existing OCR methods on the IAM dataset.