yolov3 optimized by model optimizer fail to do inference

asked 2019-12-19 08:49:58 -0500

Shay Weissman gravatar image

updated 2019-12-19 09:03:01 -0500

supra56 gravatar image

OpenCv4.1.1 windows 10 Inference engin R3 2019 Visual studio 2019. I run sample of opencv dnn. I have the following models:

  1. ssd mobilenet tensorflow model that do licenseplate recognition (ocr).
  2. ssd mobilenet tensorflow model optimized by mo_tf.py that do licenseplate recognition (ocr).
  3. Tiny yolo darknet model that locate license plates and classify jurisdiction.
  4. Tiny yolo darknet model optimized by mo_tf.py after conversion that locate license plates and classify jurisdiction.

The first 3 I am able to run correctly by the sample. The last one returns 2 matrices of size -1 -1. The routine static std::string outLayerType = net.getLayer(outLayesr[0])->type; returns "regionYolo" which is not supported by the sample. https://www.dropbox.com/sh/nir2yqqjd9... above is link to my projects and models.

edit retag flag offensive close merge delete