Very simple test to see if CLIP is accurate enough when it comes to Cosine Search and Semantic Search between images and texts. To use, place some images (preferrably a couple hundred or thousand) in the reference images folder and run the code. The image encoding might take a while and there are no loading bars or anything so you might need to be a little patient.

Best used in google colab or similar places just to test it out. This shouldnt really be the final product, its not exactly polished.