Model Optimizer SSD mobilenet V1 conversion error

[Meunier__Victor]

Hi everyone,

I have trained an SDD mobilenet v1 on hand detection and I'd like to convert it to IR file using the model optimizer. 

Here's the command I'm using:

mo.py --framework tf --data_type FP16 --input_model hand_inference_graph/frozen_inference_graph.pb --tensorflow_use_custom_operations_config ~/intel/computer_vision_sdk/deployment_tools/model_optimizer/extensions/front/tf/ssd_support.json --output "detection_boxes,detection_scores,num_detections" --tensorflow_object_detection_api_pipeline_config model-checkpoint/ssd_mobilenet_v1_coco.config

However, I have the following error during the process;

[ ERROR ]  Error while emitting attributes for layer FeatureExtractor/MobilenetV1/zeros/Output_0/Data__const (id = 66). It usually means that there is unsupported pattern around this node or unsupported combination of attributes.

I didn't find any similar problem online. Do you have an idea?

Victor

[Shubha_R_Intel]

Dear Victor, 

That is indeed a strange error.

I will PM you so that you can zip up the frozen model and send it to me over PM. But before I do that, can you verify that you are using 2019 R1 ?

Thanks,

Shubha

[Meunier__Victor]

Hi, thanks for your response. I was using the 2018.5.455 version. I'll install the 2019 R1 version to see if there's any difference.

[Meunier__Victor]

Installed the latest version but the error remains the same..

[Shubha_R_Intel]

Dear Meunier, Victor,

I believe you. Let me also try it and report back here.

Thanks for upgrading to OpenVino 2019 R1.

Shubha