Testing with own ONNX model: [ONNXRuntimeError] : 2 : INVALID_ARGUMENT : Invalid Feed Input Name:image_tensor:0 about object-detection-tf HOT 8 CLOSED

Thank you for your issue! I fixed the repo, check this out.

Works now, thankyou! So easy to use too, I would suggest you change line 51 of object_detection_image_onnx.py to:

font = ImageFont.truetype("arial.ttf", 22)

and place the arial.ttf font in the same folder as the script, otherwise when running on Linux it errors out looking in C: drive for a font (no C: drive on Linux)

from object-detection-tf.

Thank you for your issue! I fixed the repo, check this out.

from object-detection-tf.

Actually, it only seems to work with the example and not any other onnx file despite me changing the parts of the python script, I get the same error for:

line 124, in run
return self._sess.run(output_names, input_feed, run_options)
onnxruntime.capi.onnxruntime_pybind11_state.InvalidArgument: [ONNXRuntimeError] : 2 : INVALID_ARGUMENT : Invalid Feed Input Name:image_tensor:0

You can access my test onnx here: https://drive.google.com/file/d/1uY81Gh2Edh63R7G_6i5KI_oVLdMCuiem/view?usp=sharing

from object-detection-tf.

I changed the input and output names to match, now I get a common ONNX error:

[ONNXRuntimeError] : 2 : INVALID_ARGUMENT : Unexpected input data type. Actual: (N11onnxruntime17PrimitiveDataTypeIhEE) , expected: (N11onnxruntime17PrimitiveDataTypeIfEE)

from object-detection-tf.

Actually, it only seems to work with the example and not any other onnx file despite me changing the parts of the python script, I get the same error for:

line 124, in run
return self._sess.run(output_names, input_feed, run_options)
onnxruntime.capi.onnxruntime_pybind11_state.InvalidArgument: [ONNXRuntimeError] : 2 : INVALID_ARGUMENT : Invalid Feed Input Name:image_tensor:0

You can access my test onnx here: https://drive.google.com/file/d/1uY81Gh2Edh63R7G_6i5KI_oVLdMCuiem/view?usp=sharing

I looked at your archive, you are trying to run a completely different model - Yolo_v3. The thing is, my repository targets primarily on TensorFlow Detection API, not Darknet/Yolo. I can only recommend you look at the TensorRT documentation or FAQ to understand how to run your own model.

from object-detection-tf.

Thank you for your issue! I fixed the repo, check this out.

Works now, thankyou! So easy to use too, I would suggest you change line 51 of object_detection_image_onnx.py to:

font = ImageFont.truetype("arial.ttf", 22)

and place the arial.ttf font in the same folder as the script, otherwise when running on Linux it errors out looking in C: drive for a font (no C: drive on Linux)

Thank you, I'll replace it with the default python font :)

from object-detection-tf.

Actually, it only seems to work with the example and not any other onnx file despite me changing the parts of the python script, I get the same error for:
line 124, in run
return self._sess.run(output_names, input_feed, run_options)
onnxruntime.capi.onnxruntime_pybind11_state.InvalidArgument: [ONNXRuntimeError] : 2 : INVALID_ARGUMENT : Invalid Feed Input Name:image_tensor:0
You can access my test onnx here: https://drive.google.com/file/d/1uY81Gh2Edh63R7G_6i5KI_oVLdMCuiem/view?usp=sharing

I looked at your archive, you are trying to run a completely different model - Yolo_v3. The thing is, my repository targets primarily on TensorFlow Detection API, not Darknet/Yolo. I can only recommend you look at the TensorRT documentation or FAQ to understand how to run your own model.

My apologies, I was thinking that .onnx files were meant to prevent such incompatibility by producing a standardised model for running these models, I am obviously wrong! Apologies for the hassle and I hope the font fix suggestion is helpful!

from object-detection-tf.

My apologies, I was thinking that .onnx files were meant to prevent such incompatibility by producing a standardised model for running these models, I am obviously wrong! Apologies for the hassle and I hope the font fix suggestion is helpful!

It would be great if it were so :)
It's okay, take a look at the Git repositories, I'm sure you will find what you are looking for, because yolo is very popular object detection model.

from object-detection-tf.