Meme-Image-sentiment-classification-using-Machine-Learning

I employed a machine learning approach to classify the sentiment of meme images. The dataset consisted of 7000 images and was labeled as Positive, Very Positive, Negative, Very Negative, and Neutral. To address the issue of imbalance in the dataset, I consolidated certain labels and implemented data augmentation techniques for certain models. The models were trained using both text extracted from the images and image features as inputs. The final prediction was determined by taking the mode of the predictions from all of the models. To facilitate user interaction, I utilized the Flask web framework to develop a web application for the image sentiment classification.