YOLOv8 Large Pose Estimation

Our new model ZOO works with DepthAI V3. Find out more in our documentation.

0 Likes

Model Details

Model Description

YOLOv8 Pose Estimation is a YOLOv8 based convolutional neural network model designed for human pose estimation. It is highly effective and accurate even for more tricky images. The model predicts in total 17 keypoints, each representing a different part of the human body. We implement here the large version of the model.

Developed by: Ultralytics
Shared by:
Model type: Computer Vision
License:
Resources for more information:

Training Details

Training Data

is a large-scale object detection, segmentation, and captioning dataset.

Testing Details

Metrics

Results of the mAP and speed are evaluated on COCO Keypoints val2017 dataset with the input resolution of 640×640. Results are taken from .

Model	mAP (50-95)	Params ^(M)
	67.6	44.4

Technical Specifications

Input/Output Details

Input:
- Name: images
  - Info: BGR image
Output:
- Name: multiple (see NN archive)
  - Info: Classification scores, bounding boxes, and keypoints for a multitude of detections.

Model Architecture

Backbone: CSPDarknet53
Head: Anchor-free object detection head

Consult the for more information.

Throughput

Model variant: yolov8-large-pose-estimation:coco-640x352

• Input shape: [1, 3, 352, 640] • Output shapes: [[1, 6, 44, 80], [1, 6, 22, 40], [1, 6, 11, 20], [1, 51, 3520], [1, 51, 880], [1, 51, 220]]

• Params (M): 44.479 • GFLOPs: 47.767

Platform	Precision	Throughput (infs/sec)	Power Consumption (W)
RVC4	INT8	168.82	6.34

* Benchmarked with , using 2 threads (and the DSP runtime in balanced mode for RVC4).

* Parameters and FLOPs are obtained from the package.

Quantization

The RVC4 version of the model was quantized using a custom dataset. This was created by taking 40 images containing people from the dataset.

Utilization

Models converted for RVC Platforms can be used for inference on OAK devices. DepthAI pipelines are used to define the information flow linking the device, inference model, and the output parser (as defined in model head(s)). Below, we present the most crucial utilization steps for the particular model. Please consult the docs for more information.

Install DAIv3 and depthai-nodes libraries:

pip install depthai
pip install depthai-nodes

Define model:

model_description = dai.NNModelDescription(
    "luxonis/yolov8-large-pose-estimation:coco-640x352"
)

nn = pipeline.create(ParsingNeuralNetwork).build(
    <CameraNode>, model_description
)

Inspect model head(s):

YOLOExtendedParser that outputs message (bounding boxes and body pose keypoints of detected people).

Get parsed output(s):

while pipeline.isRuning():
    parser_output: ImgDetectionsExtended = parser_output_queue.get()

Example

You can quickly run the model using our script. It automatically downloads the model, creates a DepthAI pipeline, runs the inference, and displays the results using our DepthAI visualizer tool. To try it out, run:

python3 main.py \
    --model luxonis/yolov8-large-pose-estimation:coco-640x352

The large version of YOLOv8 Pose Estimation model for human pose estimation.
License	GNU Affero General Public License v3.0 Commercial use
Downloads	113
Tasks	Keypoint Detection
Model Types	ONNX

Model Variants

Name	Version	Available For	Created At	Deploy
		RVC4	10 months ago