You can deploy this model and start earning money today!

LibreYOLO/
LibreRFDETRs-pose

Name: LibreRFDETRs-pose
Availability: InStock

detection-datasets/coco

apache-2.0

Views

No views yet

LibreRFDETRs-pose

EXTREMELY experimental RF-DETR-small pose checkpoint for LibreYOLO.

This is a COCO-17 human pose checkpoint trained for early validation of LibreYOLO's RF-DETR task="pose" path. It should be treated as experimental while the pose implementation and release process settle.

Checkpoint

File: LibreRFDETRs-pose.pt
Family: LibreRFDETR
Size: s
Task: pose
Keypoints: COCO-17, (x, y, visibility)
Selected input size during final validation: 768

COCO Keypoint Validation

Validation was run on COCO person keypoints val2017 through LibreYOLO's pose validator.

Metric	Value
keypoints mAP50-95	`0.604812`
keypoints mAP50	`0.866018`
keypoints mAP75	`0.669` from COCO summary
keypoints AR50-95	`0.713` from COCO summary

The selected checkpoint is from the first epoch of the final 768x768, batch-10 refinement phase. The training history is staged continuation from an RF-DETR-small detection backbone, with 18 logged validation epochs across the 512, 640, and 768 refinement phases.

Usage

python

from libreyolo import LibreRFDETR

model = LibreRFDETR("LibreRFDETRs-pose.pt", task="pose")
results = model.predict("image.jpg")

Caveats

Experimental checkpoint, not a final benchmark release.
Pose export/runtime backends may have separate support status from PyTorch inference.
Metrics are from the LibreYOLO training run artifacts in PR development, not from an independent external benchmark suite.

LibreYOLO/LibreRFDETRs-pose