Holistic Features are almost Sufficient for Text-to-Video Retrieval | Read Paper on Bytez