YOLOv8-Pose
YOLO with a keypoint head: detection, tracking, and pose in one pass, from nano to x-p6.
About
YOLOv8-Pose bolts a keypoint head onto Ultralytics' detector, so one forward pass gives bounding boxes plus the 17 COCO keypoints. The family spans yolov8n-pose.pt (3.3M params, genuinely real-time on CPU) up to yolov8x-pose-p6, and the same package handles tracking through ByteTrack or BoT-SORT.
What it lacks in peak accuracy it makes up for in ecosystem: pip install ultralytics, export to ONNX/TensorRT/CoreML, and annotation/training tooling through the Ultralytics Platform. The catch is the license — the weights and code are AGPL-3.0, so commercial products need the paid Enterprise license.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Provider-reported (Ultralytics docs, single-model single-scale COCO val2017): yolov8n-pose 50.4, yolov8s-pose 60.0, yolov8m-pose 65.0, yolov8l-pose 67.6, yolov8x-pose 69.2, yolov8x-pose-p6 71.6 mAP50-95. No independent published benchmarks were found.
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 7/10
Scored 7/10 (metrics, metrics-v1). Evidence: official Ultralytics release with published COCO metrics and downloadable weights; enormous adoption across real-time pose projects. Strengths: weights downloadable, excellent tooling and export story, reproducible metrics. Docked for: (1) AGPL-3.0 is strongly copyleft — commercial users must pay, a major openness hit; (2) all accuracy figures are provider-reported. License: AGPL-3.0; commercial via Ultralytics Enterprise.
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 10.0/10
- Maintenance · recency, releases, issue load
- 9.0/10
- Openness · license + docs
- 4.0/10
- Manual review · sanity read + red-flag check
- 7.0/10
- · Ultralytics (2023); single-shot 17-keypoint pose from yolov8n-pose (3.3M params) to yolov8x-pose-p6.
- · Provider-reported COCO val mAP50-95: 50.4 (n) → 69.2 (x) → 71.6 (x-p6). No independent benchmarks.
- · Tracking via ByteTrack/BoT-SORT; export to ONNX/TensorRT/CoreML.