Closed WanliZhong closed 2 weeks ago
I should config the model, my mistake. But the preprocess parameters are still different from the official doc. Should it change?
The images must be loaded in RGB with a range of [0, 1] per channel, then normalized per-image using mean = [0.485, 0.456, 0.406] and std = [0.229, 0.224, 0.225].
System Information
OpenCV version: 4.9.0.80 Operating System / Platform: MacOS 14.4.1
Detailed description
I got results:
src:
BTW: the model file use different scale and rgb channel with the original document form onnx model zoo
https://github.com/opencv/opencv/blob/05a54b1405bdd7ef5d7cf3d3215332cac50f0416/samples/dnn/models.yml#L272-L282 relate PR: https://github.com/opencv/opencv/pull/24397
Steps to reproduce
Just use the sample in https://github.com/opencv/opencv/blob/4.x/samples/dnn/segmentation.py
python segmentation.py --model path/to./fcn-resnet101-11.onnx --input path/to/dog_orig_size.png
Issue submission checklist