Yolov5s onnx model inference

[anazkhan]

Search before asking

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

Question

Hi , I am unable to get bounding boxes from the output i got by running yolov5s onnx model in onnx runtime. The output is list of arrays of the shape (3,52,52,85) , (3,26,26,85) , (3,13,13,85) respectively . it will be helpful if you can provide me with the postprocess code to define the bounding boxes.

Additional

No response

[UltralyticsAssistant]

👋 Hello @anazkhan, thank you for your interest in YOLOv5 🚀! Please check out our ⭐️ Tutorials for guidance on various tasks such as ONNX Export and Inference.

If this is a 🐛 Bug Report, please provide a minimum reproducible example to help us debug it. For your specific issue, ensuring you have the correct post-processing steps is key when working with ONNX outputs.

Also, make sure you meet the following requirements:

Requirements

Python>=3.8.0 with all requirements.txt installed. 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 any of the following environments:

• Notebooks with free GPU:
• Google Cloud, AWS, Docker: See respective Quickstart Guides

Status

Check our CI Status:

If the badge is green, all tests are passing 👍.

Introducing YOLOv8 🚀

Explore our state-of-the-art YOLOv8 here for enhanced capabilities in object detection and image processing tasks. Install with:

pip install ultralytics

This is an automated response. An Ultralytics engineer will assist you further soon. Thanks for your patience! 😊

[pderrenger]

@anazkhan to extract bounding boxes from the YOLOv5s ONNX model output, you'll need to apply non-max suppression and decode the predictions. You can refer to the post-processing steps in the YOLOv5 repository's detect.py script, which includes functions for these tasks. If you need further guidance, please check the YOLOv5 documentation for detailed instructions.