berak

please make sure you give a proper image path (DOUBLE CHECK!) to the `--back` [argument of the script](https://github.com/microsoft/ailab/blob/3bfc475eef00232fce655d94199cc20c7a53d573/VirtualStage/BackgroundMatting/background_matting_image.py#L355)

@Henrilux -- you probably misunderstood something, it's NOT AT ALL about installing opencv. you have to invoke the python script with a proper `--back /some/valid/imagepath` argument, that's all.

already the 1st createTrackbar() calls Refresh() (since there is a value pointer), and the other two bars are not yet created: ``` Thread 1 hit Breakpoint 1, Refresh (data=0x64fad0) at...

@wwdok , maybe dont use callback functions, but a while loop, where you poll the current trackbar values

iirc it is from this repo: https://github.com/WongKinYiu/yolor and the model is here: !wget "https://drive.google.com/uc?export=download&id=1Tdn3yqpZ79X7R1Ql0zNlNScB1Dv9Fp76&confirm=1" -O yolor_p6.pt

`borderValue` is indeed a `cv::Scalar`, so for non 0 values (the default), you need to initialize all 4 components of it like borderValue = (np.nan, np.nan, np.nan, np.nan) # borderValue...

tried with model/code [from here](https://github.com/WongKinYiu/yolov7) on colab: ``` !git clone https://github.com/WongKinYiu/yolov7 %cd yolov7 !wget https://github.com/WongKinYiu/yolov7/releases/download/v0.1/yolov7.pt from models.experimental import attempt_load model = attempt_load("yolov7.pt", map_location="cpu") # load FP32 model model.eval() import torch...

@wwzh2015 > Now yolov7 project no support export to onnx format it does ! https://github.com/WongKinYiu/yolov7/blob/main/models/export.py update: @fengyuentau , thanks for the nice link ! there's a [resources.tar.gz file here](https://github.com/hpc203/yolov7-opencv-onnxrun-cpp-py/issues/2#issuecomment-1186215405) with...

@hermano360 no. it's purpose is to *detect* faces. what you want is called *face recognition*

great project !!