This work is about predicting whether the given image or text is spam or not.
The project comprises of two spam detection tasks: text spam detection and image spam detection.
For text-based spam detection, SVC achieves an accuracy of 99.15% whereas, a DNN based model is proposed for image spam detection attains an accuracy of 95.83%. The ensemble of both the models by providing a weight of 2:1 for image:text provides an accuracy of 96.93%.
Email: [email protected]