The rapid development of consumer products with short life spans, along with fast, global e-commerce and e-marketing distribution of products and services requires greater corporate diligence to protect intangible assets such as brands which can easily be coped or placed in grey markets. Trademarks are the government registered legal intellectual property rights (IPRs) used to protect companies’ brands and build brand equity. Given the rapid growth in the number of global trademark registrations, the number of trademark infringement cases is also increasing, a great challenge for the original trademark owner to detect the infringement and takes action to protect the brand image and related commercial interests. This research develops a trademark similarity assessment methodology based on the US trademark law related to the high likelihood of confusion and associated regulations. The research focuses on identifying trade mark image similarity using a deep learning approach. The convolutional neural network (CNN) and Siamese neural network (SNN) algorithms are modeled and trained using Cifar-10 and TopLogo-10 corpuses. These corpuses consist of more than 100,000 positive image pairs and more than 150,000 negative image pairs as training data. After training the model, an image input to the model extracts and recommends similar trade mark images found in the corpus. The solution assists users registering new trademarks to identifying similar marks that may lead to disputes. The solution also automatically screens images to identify marks that potentially infringe upon registered trademarks.