Details for Output Tensor

[ShreyPatel1311]

Search before asking

• I have searched the YOLOv5 issues and discussions and found no similar questions.

Question

Pose Landmark coordinates' Range in output tensor

Additional

Hello guyz,
first of all great work with YOLOv5, I am able to use it in my embedded systems without any lags. The output size for this model is shown to be (1, 56, 8400). I understood that there are 8400 predictions made by the model. What I would like to know is about 56 features which the model is predicting. Mainly, I would like to get the coordinates for the human-pose detected by model. Can anyone please tell me what is the range for pose landmarks?!

Regards

[UltralyticsAssistant]

👋 Hello @ShreyPatel1311, thank you for your interest in YOLOv5 🚀 and for your feedback! Your question about pose landmark coordinates in the output tensor is a good one.

This is an automated response to help you get started while an Ultralytics engineer reviews your issue and will assist you soon 😊.

For questions like this, it’s helpful if you can provide a minimum reproducible example (MRE) that demonstrates your workflow and the specific output you are seeing. This allows us to better understand your setup and provide more precise guidance.

Please also make sure you are following our Tips for Best Training Results and include as much detail as possible, such as example images, code snippets, and any logs if applicable.

Requirements

Python>=3.8.0 with all requirements.txt installed including PyTorch>=1.8. To get started:

git clone https://github.com/ultralytics/yolov5  # clone
cd yolov5
pip install -r requirements.txt  # install

Environments

YOLOv5 can be run in these verified environments (with all dependencies including CUDA/CUDNN, Python and PyTorch preinstalled):

• Notebooks with free GPU:
• Google Cloud Deep Learning VM. See GCP Quickstart Guide
• Amazon Deep Learning AMI. See AWS Quickstart Guide
• Docker Image. See Docker Quickstart Guide

Status

If this badge is green, all YOLOv5 GitHub Actions Continuous Integration (CI) tests are currently passing. CI tests verify correct operation of YOLOv5 training, validation, inference, export and benchmarks on macOS, Windows, and Ubuntu every 24 hours and on every commit.

Thanks again for your question, and an Ultralytics engineer will reply with more details soon!

[pderrenger]

The 56 features in your YOLOv5 pose model output represent: 4 bounding box coordinates (x, y, w, h) + 1 object confidence + 51 pose keypoint values (17 keypoints × 3 values each for x, y, visibility). The pose landmark coordinates are typically normalized to the input image dimensions, so x and y coordinates range from 0.0 to 1.0, while visibility scores range from 0.0 to 1.0 indicating keypoint confidence.