-2

I trained custom object detector with loss 2.x , when i try to export it i am getting the following error

I came across this : 'Parsing Inputs... Incomplete shape' error while exporting the inference graph in Tensorflow

But Iam sure label is same : I ran it multiple times

python3 export_inference_graph.py --input_type image_tensor       --pipeline_config_path training/inception_v2.config --trained_checkpoint_prefix training/model.ckpt-688 --output_directory trained-inference-graphs/output_inference_graph_v1

I'm getting the error :

114 ops no flops stats due to incomplete shapes. Parsing Inputs... Incomplete shape.

For complete report please check : https://pastebin.com/mGSBDgJC

Edit : The model is detecting with eval using python3 eval.py --logtostderr --pipeline_config_path=/home/ic/Documents/objectExtraction/workspace/training_demo/training/inception_v2.config --checkpoint_dir=/home/ic/Documents/objectExtraction/workspace/training_demo/training --eval_dir=/home/ic/Documents/objectExtraction/workspace/training_demo/eval

anilkunchalaece
  • 388
  • 2
  • 10

1 Answers1

1

This happens since you didn't state what is the input resolution with the argument input_shape. It simply means that it cannot compute how many flops an operation will take, since it doesn't know what the input resolution will be. You can still use the exported graph for inference without problem.

netanel-sam
  • 1,862
  • 9
  • 14
  • Hi, Thanks. I'm able to use the graph by ignoring the warning. input resolution means image resolution ? – anilkunchalaece Feb 07 '19 at 06:03
  • Yes, You can read the help of `export_inference_graph.py`: input_shape: If input_type is `image_tensor`, this can explicitly set the shape of this input tensor to a fixed size. The dimensions are to be provided as a comma-separated list of integers. A value of -1 can be used for unknown dimensions. If not specified, for an `image_tensor, the default shape will be partially specified as `[None, None, None, 3]`. – netanel-sam Feb 07 '19 at 08:28