YOLO26 Object Detection (ONNX)

[tbc725]

This is taken from “YOLO26 Object Detection #355” and has been modified from using .pt files to using .onnx files.

Important

• The user will need to download or convert the YOLO26 ONNX models manually (the Nano ONNX model is provided in the download).
• The user will need to copy the “bin” folder from the YOLOv8 modules and then update their Ultralytics module to the latest version. See (YOLO26 Object Detection #355 (comment))
• The user will need to install the onnxruntime python module to its venv.
• CPU only; GPU not tested.
• Windows only; not tested on other platforms.

Note: The .onnx model size is larger than the .pt format, but based on testing, it runs almost twice as fast as the .pt format on the CPU.

Skeleton-ObjectDetectionYOLO26onnx.zip