object-detection-opencv icon indicating copy to clipboard operation
object-detection-opencv copied to clipboard

opencv cannot read net - Error of "Unknown layer type"

Open chenharel1 opened this issue 5 years ago • 0 comments

Hello, I am trying to use your code for running detection of yolo v1. I use python 3.5 I use opencv 4.1.0 The error occurs in line 57: net = cv2.dnn.readNet(args.weights, args.config) It says:

cv2.error: OpenCV(4.1.0) /io/opencv/modules/dnn/src/darknet/darknet_io.cpp:668: error: (-212:Parsing error) Unknown layer type: connected in function 'ReadDarknetFromCfgStream'

I took the weights and config from the original Darknet. I tried weights and cfg for yolov1, as well as for yolov1 tiny, but the same error happens for both. On the other hand, when I tried weights and cfg for yolov3, it succeeded. So, does cv2.dnn modules are not compatible with yolo1? Is that the problem or anything else?

Thanks!

chenharel1 avatar Jun 05 '19 14:06 chenharel1