DForged-Net: Discrimination Among Original, Forged, and AI -Generated Handwritings

A document is handwritten, printed, or electronic material that contains information or evidence and serves as an official or personal record. Handwritten documents often undergo modifications, such as additions and alterations with similar inks, by AI -generated or humans, but these changes are challenging to identify or perceive with the human eye. Hence, there is a requirement for techniques that can automatically identify AI -generated or forged handwriting in order to achieve precise detection efficiency. Therefore, our proposed DForged-Net model will discriminate between several types of fake handwriting and original handwriting in the modified scanned document. Deep learning techniques are employed during decision-making, wherein a fresh set of characteristics is extracted for individual document words. The experimental results demonstrate that the proposed method exhibits a high level of proficiency in detecting and localizing all types of text classes, such as Authentic Hand-writing, Printed Texts, AI or Machine Generated Handwriting, Forged Handwriting, and Non-texts in modified handwritten documents.