Fast Yolov3 CPU Inference Using Onnxruntime

[matankley]

Hi,
I would like to thank you and share with you few insights I observed while testing inference of Yolov3 Object detection models using opencv-dnn and onnxruntime on CPU.

Onnxruntime inference is significantly faster than the common opencv-dnn and darknet that are usually used for yolov3 models.

Here is a link to the full article - (https://medium.com/towards-artificial-intelligence/yolov3-cpu-inference-performance-comparison-onnx-opencv-darknet-6764f2bde33e)

Hope this can help you achieve reasonable performance on low-powered devices.