 This research explored the use of convolutional neural networks, CNNs, and support vector machines, SVMs, to detect pornography in movies. Three different CNN architectures, MobileNet, VGG, 19, and ResNet 50 underscore V2, were used to identify frames for movies as either pornographic or not. These models were then combined with SVMs to improve the accuracy of the results. The best accuracy of 92.8% was achieved when the ResNet 50 underscore V2 architecture was used as the feature extractor and SVM as the classifier. By transferring knowledge from other similar tasks, this research has shown how machine learning techniques can be used to detect pornography in movies more accurately than manual censorship methods. This article was authored by Hor Sway Lin, Serena Manser, Neuer Aldehal, and others.