Datasets for SketchVLM: Vision-Language Models Can Annotate Images to Explain Thoughts and Guide Users (https://sketchvlm.github.io/)