OpenCV dnn MobileNet v2 support

asked 2018-04-05 09:52:35 -0600

piojanu gravatar image

Hi!

Is MobileNet v2 supported? I've exported one from my TF Object Detection API training (I fallowed instruction on your site and I was able to successfully export MobileNet v1 before) and I get following error:

cv2.error: /io/opencv/modules/dnn/src/tensorflow/tf_importer.cpp:553: error: (-2) Input layer not found: FeatureExtractor/MobilenetV2/expanded_conv_13/expand/weights in function connect

If it's not supported, are there plans to implement missing layer?

Thanks!

edit retag flag offensive close merge delete

Comments

I am having a similar issue with SSD MobileNet v2 and OpenCV. I have trained ssd_mobilenet_v2_coco.config in TensorFlow 1.15.0 and exported the frozen graph (.pb and .pbtxt). OpenCV is giving the following error at forward pass:

cv2.error: OpenCV(4.2.0) C:\projects\opencv-python\opencv\modules\dnn\src\dnn.cpp:562: error: (-2:Unspecified error) Can't create layer "FeatureExtractor/MobilenetV2/expanded_conv_2/add" of type "AddV2" in function 'cv::dnn::dnn4_v20191202::LayerData::getLayerInstance'

I am using OpenCV 4.2.0.32, Windows 10, Python 3.6.8

sigmoid gravatar imagesigmoid ( 2020-03-25 08:23:02 -0600 )edit
add a comment