Convert TensorFlow Models Problem

[zhao__ying]

Hi:

Convert IR result find Log level is Error。this info  The Preprocessor block has been removed. Only nodes performing mean value subtraction and scaling (if applicable) are kept. Does this log represent a very serious mistake?

Convert IR out . and run object_detection_sample_ssd progress ,but the result is wrong. It's discrimination is lower than 0.07。but this pb is converted from tensorflow which discrimination is higher than 0.5。

I don't konw  how to solve this problem and even don't kown how to find the error？

  Pls show me the way to solve this problem. Thanks.

my  command:

 ./mo_tf.py  --input_model ./tmod-tax-1219-31000.pb --tensorflow_use_custom_operations_config ./extensions/front/tf/ssd_v2_support.json --tensorflow_object_detection_api_pipeline_config ./ssd_mobilenet_v1_coco.config --data_type=FP16 --output_dir out

[zhao__ying]

[zhao__ying]

Anybody can help me solve this problem?

[whatthisismyname]

You may want to share what model are you training and the command you are using. it worked fine for me on SS_MobileNET and SSD_inceptionV2.