Intel® Distribution of OpenVINO™ Toolkit
Community support and discussions about the Intel® Distribution of OpenVINO™ toolkit, OpenCV, and all things computer vision-related on Intel® platforms.

Unable to convert FaceNet tensorflow model to IR using modeloptimizer

saif_s_
Beginner
509 Views

The frozen graph model used is attached and when I try to summarize the graph, I could see 2 inputs (summary_graph.png). When I try to use the modeloptimizer to create IR it is failing(ir_convert_error.png). The command I used to convert the FaceNet model to IR is as follows.

 

python3 modeloptimizer/scripts/model_optimizer.py \

--input_model=/home/saif/Downloads/20170511-185253.pb \

--input=input:0 \

--output=embeddings \

--data_type=float  

--input_shape 1,160,60,3 \

--model_name InceptionV1

 

Need help in understanding the reason of this and also what necessary changes needs to be done for correct conversion to IR

0 Kudos
2 Replies
Jeffrey_M_Intel1
Employee
509 Views

Just wanted to let you know that improvements for Facenet have been added which will be available in the next release.

Mickael_T
Beginner
509 Views

Hi Jeffrey, Thank you very much for following up on this. I am currently on OpenVINO 2018.2.319. I will try again to generate the Intermediate Representation with Model Optimizer.

Reply