Improving Students' Understanding of Education Using Generative Adversarial Networks: Bridging Text and Visuals

Utilizing technology in the classroom is essential in today's environment to enhance students' learning. The background problem revolves around the necessity to adapt traditional educational methods to meet the evolving needs of students in today's digital age. With rapid technological advancements, classrooms face the challenge of effectively integrating technology to enhance learning experiences and outcomes. This includes addressing issues such as access to technology, digital literacy among students and educators, and the effective utilization of technology to facilitate interactive and engaging learning environments. The need to employ Generative Model to link school students descriptive ‘Poem’ statements with relevant visual imagery, accommodating diverse learning preferences. Through pre-trained VQGAN and CLIP models, this research work seamlessly integrates textual description with visually captivating images, employing iterative optimization methods to align image representations with textual descriptions. By incorporating advanced technique such as cutout augmentation the project ensures the quality and variety of generated images, catering to the diverse learning needs of school students. The outcome is successfully produced and an image is generated matching the ‘Poem’ description, where the average approximate percentage of accuracy is approximately 77.80^% and the average approximate percentage of loss is approximately 22.89%. Additionally, a video documenting the iterative image synthesis process was created, providing insights into the image's evolution and highlighting the effective integration of technology to enhance educational experiences.