Currently, recall doesn't parse any info from images on web pages or documents. This could be changed by using Llama Parse API https://docs.cloud.llamaindex.ai/llamaparse/getting_started