Inference with OpenCV

[JonathanSamelson]

Hi,

I wanted to use your caffe model trained on DETRAC with OpenCV to do inference on a test image but I get the following error:

cv2.error: OpenCV(4.5.1) C:\Users\samelson\AppData\Local\Temp\pip-req-build-fupgqc0e\opencv\modules\dnn\src\dnn.cpp:3113: error: (-215:Assertion failed) inp.total() in function 'cv::dnn::dnn4_v20201117::Net::Impl::allocateLayers

I tried with both vgg16_eb_1-3-5_final.caffemodel and vgg16_eb_5_final.caffemodel

Here is the code I used:

    model_path = "vgg16_eb_1-3-5_final.caffemodel"
    config_path = "test_1-3-5.prototxt.txt"

    frame = "path/to/test/image.jpg"

    net = cv2.dnn.readNet(model_path, config_path)
# or 
   net = cv2.dnn.readNetFromCaffe(config_path, model_path)

    image = cv2.imread(frame)
    blob = cv2.dnn.blobFromImage(image, 1, (input_width, input_height), (104, 117, 123))

    net.setInput(blob)
    outputs = net.forward()

Does anyone know if it's possible to do inference with OpenCV with those models or is there particular layers that are not implemented in OCV?

Thank you in advance.