Handwritten examination evaluation using LLMs and GenAI

After completion of an examination, students usually find it difficult to wait for their result. Traditionally, paper checking is considered time-consuming and not that effective seeing that it is checked by humans who tend to make mistakes. The amount of energy spent on reading and understanding the subjective response is difficult and may result is NOT an accurate result. This research paper introduces an Automated model which is designed using Python and incorporates ML Techniques like Bidirectional Encoder Representations from Transformers (BERT), Sentence-Bidirectional Encoder Representations from Transformers (SBERT), Term Frequency-Inverse Document Frequency (TF-IDF), and Natural Language Processing (NLP). For recognizing handwritten text, the Google Vision API and for dynamic comparison with model answers, the Gemini API helps generate similarity scores. The system also supports cloud-based result storage, enabling long-term tracking of student progress and minimizing the risks of data loss associated with traditional methods.