I utilized the openai clip model to create a classifier for the (beauty) and (fashion) categories, and it achieved 100% accuracy. However, this level of accuracy may not always be possible due to the limited number of provided images. To overcome this limitation, I had to augment the dataset to generate more images.