Psychological Analysis of Social Media Visual Content Based on Image Recognition Algorithm

Technological advancement has seen social media airways availing a steady stream of visual content, which has compounded the need to create new and effective ways of uncovering the psychological aspects inherent to these images. This work aims to investigate the possibility of applying modern image processing techniques to classify the affective and referential features of social media images. To analyze and feature extract from a diverse set of images embracing social media applications, the study undertook the application of a Convolutional Neural Network (CNN), which was refined through transfer learning. The model’s performance was confirmed by its high accuracy levels of 92% for the thematic categories and 88% for the emotional content, together with Pearson coefficients for the comparison with human-coded benchmarks at r= 0. 84 for the emotional scores and r= 0. 79 for the thematic scores. The results suggest that such an approach is capable of recreating human decision-making processes with reasonable accuracy in the context of SEM interpretation relevant to digital media literacy, mental health, and marketing techniques. It also discusses the problems connected to ethical issues, confidentiality and anonymity of participants’ data preventing algorithmic bias and making it fair and inclusive in analyzing the data of the study. We consider certain limitations like selection bias and subjectivity of annotations; therefore, it suggests the future directions of the research like using larger scale and different types of data sets as well as using multimodal data. Therefore, this research validates the approach to synchronising image recognition technologies with psychological analysis theories for unpacking the complex psychological aspects of social media visuals. Thus, these methodologies will also be vital for the further advancement of research in the field of modern digital communication and the investigation of people’s interactions and emotions in this context.