search

lindevs / yolov8-face

Pre-trained YOLOv8-Face models.
24 stars 4 forks source link

Regarding the output issue of ONNX model #2

Closed zhaoyangwei123 closed 1 week ago

zhaoyangwei123 commented 2 weeks ago

Hello author, thank you very much for your open source work. However, when using onnx for facial recognition inference, I need to use the output of the YOLOV8 facial detection model as input for subsequent models. However, through the onnx viewer, I observed that the output of the facial detection model is 1 80 20 * 20. May I ask why it is not the (x, y, w, h) coordinates? Facial detection should obtain the position of the detection box.

lindevs commented 1 week ago

Hi. The ONNX models available on the releases page produce an output with the shape [1, 5, 8400]. You can use the Netron application to inspect the model structure. For processing the model output and extracting the (x, y, w, h) coordinates, refer to the OpenCV DNN example.