The classification of images is a difficult problem in pattern recognition and computer vision. Unwanted public content is known as spam, while unwanted content included in photographs is known as image spam. Systems for email communication are at risk from image spam. There is a lot of unwanted content online today all across the Internet. Spam can easily circumvent these text based spam detection systems, despite the fact that many Text-based spam can be successfully detected using machine learning algorithms. In this project, the deep learning detection methods are studied, evaluated, and compared with pertinent existing tools like SVM, Gaussian Techniques and PCA. In our study, a new tool for detecting picture spam is introduced, and it is contrasted with current techniques.