Mood Detection using CNN is a multiclass image classification project aimed at classifying images of people into four different emotional states: Happy, Sad, Angry, and Scared. The project achieves an impressive accuracy of approximately 99 percent. The dataset used for training the model was built from scratch by downloading images of individuals exhibiting the aforementioned emotional states from the web. These images were then arranged into folders and any questionable or low-quality photos were removed to enhance the training process.
The dataset used for training the model was curated from various sources on the internet. Images were collected depicting individuals displaying happy, sad, angry, and scared facial expressions. These images were meticulously arranged into separate folders representing each class.
The Convolutional Neural Network (CNN) architecture used for this project was designed to effectively learn and classify images based on their emotional content. The architecture consists of multiple convolutional layers followed by max-pooling layers to extract features from the input images. This is followed by fully connected layers and softmax activation for the final classification.