Skip to content
B bounding-box Point the LLM at a box

Point the LLM at a bounding box.

Upload a PDF or image, say what you want found, and a vision model draws the box around it.

PDF or image (.pdf, .png, .jpg, .webp, .bmp…)

Say “all” or a plural to detect multiple objects (e.g. “all circles”). For a multi-page PDF, pick the page above — only that page is sent to the model.

Your image is normalized (long edge capped at 1280px) before being sent to the model; nothing is stored beyond the detection record and the normalized copy.

Recent

View all →