Some object detection methods doesn't detect anything

asked 2019-04-16 12:52:58 -0500

Jesuino Vieira gravatar image

Hello everybody,

I'm trying to test opencv tested dnn models. I downloaded all config and weights files from this code and got the blobFromImage input parameters for each method here. Also, I have based all my codes on this example.

I'm note sure if the parameters are correctly, the following methods didn't detect anything:

  • Tensor flow

ssd-inceptionv2-coco

ssd-mobilenetv1-coco

ssd-mobilenetv2-coco

  • Caffe

faster-rcnn-vgg16-voc

rfcn-resnet50-voc

ssd-vgg16-voc

  • Yolo

yolov2

tiny-yolov2

  • ONNX

tiny-yolov2

I made the codes available in my github. I'm not sure if the problem is in the parameters, in the manipulation of the vector that the forward method returns or something else. I'll be glad if somebody can help me. Thanks in advance.

Obs.: it was not possible upload the weight files because they are too big, but they are all in opencv repo.

edit retag flag offensive close merge delete