rf_detr_detect_objects – 1 MCP server offers this tool
What it usually takes
checkpoint– in 1 of 1confidence– in 1 of 1input_image– in 1 of 1resolution– in 1 of 1
No server marks it as only reading: treat it as able to change data.
From each publisher's own tool list. We never call a tool.
Servers
| Server | Inputs | Read-only? |
|---|---|---|
RF-DETR co.huggingface.gf-john/rf-detr Detect objects in an image using RF-DETR and return the annotated image plus structured detections. RF-DETR is a real-time, transformer-based object detector; this tool runs the COCO-pretrained checkpoints, so it recognizes the 80 standard COCO classes (person, car, dog, ...). It produces two outputs. Output 1 is the input image with a labeled bounding box drawn over every detected object. Output 2 is a JSON list of detections, one entry per object, where each entry has these fields: "class_id" (integer COCO class index, 0-79); "classname" (string, the human-readable COCO class name for that class_id); "confidence" (float detection score between 0.0 and 1.0); and "bounding_box". The "bounding_box" field is a list of four numbers [x_min, y_min, x_max, y_max] giving the top-left corner (x_min, y_min) and bottom-right corner (x_max, y_max) in ABSOLUTE PIXEL coordinates of the ORIGINAL input image (not the inference resolution). This is the albumentations "pascal_voc" bounding-box format, | checkpoint,confidence,input_image,resolution | Not stated |