Detecting Harmful Memes and Their Targets | Papers With Code