When should I set the options tensorflow_use_custom_operations_config and tensorflow_object_detection_api_pipeline_config?

[es__we]

I have coverted some object detection models to IR format, in these cases, I do as follows:

python3 /opt/intel/openvino/deployment_tools/model_optimizer/mo_tf.py \
--input_model frozen_inference_graph.pb \
--data_type FP16 \
--reverse_input_channels \
--batch 1 \
--tensorflow_use_custom_operations_config  /opt/intel/deployment_tools/model_optimizer/extensions/front/tf/ssd_v2_support.json \
--tensorflow_object_detection_api_pipeline_config pipeline.config

It works, however, I also found that someone doesn't set these two parameters. What's more, I didn't modify the file path in pipline.config such as:

train_input_reader {
  label_map_path: "PATH_TO_BE_CONFIGURED/mscoco_label_map.pbtxt"
  tf_record_input_reader {
    input_path: "PATH_TO_BE_CONFIGURED/mscoco_train.record"
  }

Doesn't it matter?

[JesusE_Intel]

Hi we,

Thanks for reaching out. The paths to be configured are only used when training the model. When using the Model Optimizer, you only need to specify the pipeline.config with the --tensorflow_object_detection_api_pipeline_config as you mentioned. 

Please let me know if this answers your question.

Regards,

Jesus

[es__we]

Thanks, I see, I feel the need to learn more about the document.