Problem with Onnxruntime

[nldhuyen0047]

I could not run the code "git checkout liqun/ImageDecoder-cuda" because there is not the directory liqun/ImageDecode-cuda, could you help me please with this problem?

[warmshao]

I could not run the code "git checkout liqun/ImageDecoder-cuda" because there is not the directory liqun/ImageDecode-cuda, could you help me please with this problem?

cd onnxruntime？then git checkout

[nldhuyen0047]

yes, i did that and i checked the github of onnxruntime, is there any problems with liqun branch

[warmshao]

yes, i did that and i checked the github of onnxruntime, is there any problems with liqun branch

liqun/ImageDecoder-cuda is branch，use git branch to verify

[nldhuyen0047]

thank you so much, i got it

[nldhuyen0047]

but i have had this error, do you know about it: onnxruntime.capi.onnxruntime_pybind11_state.InvalidArgument: [ONNXRuntimeError] : 2 : INVALID_ARGUMENT : Non-zero status code returned while running GridSample node. Name:'/dense_motion_network/GridSample' Status Message: Only 4-D tensor is supported

[warmshao]

but i have had this error, do you know about it: onnxruntime.capi.onnxruntime_pybind11_state.InvalidArgument: [ONNXRuntimeError] : 2 : INVALID_ARGUMENT : Non-zero status code returned while running GridSample node. Name:'/dense_motion_network/GridSample' Status Message: Only 4-D tensor is supported

Tell me about your runtime environment and how you compile onnxruntime?

[nldhuyen0047]

I created a virtual environment and installed packages. I followed what you instructed at "ONNX Inference" to run by GPU on Linux

[warmshao]

I created a virtual environment and installed packages. I followed what you instructed at "ONNX Inference" to run by GPU on Linu

I created a virtual environment and installed packages. I followed what you instructed at "ONNX Inference" to run by GPU on Linux

How about using Docker? It's much simpler that way

[nldhuyen0047]

Sample node. Name:'/dense_motion_network/GridSample' Status Message: Only 4-D tensor is supported Traceback (most recent call last): File "/home/code/FasterLivePortrait/run.py", line 63, in dri_crop, out_crop, out_org = pipe.run(frame, img_src) File "/home/code/FasterLivePortrait/src/pipelines/faster_live_portrait_pipeline.py", line 329, in run out_crop = self.model_dict["warping_spade"].predict(f_s, x_s, x_d_i_new) File "/home/code/FasterLivePortrait/src/models/warping_spade_model.py", line 34, in predict preds = self.predictor.predict(feature_3d, kp_driving, kp_source) File "/home/code/FasterLivePortrait/src/models/predictor.py", line 211, in predict results = self.onnx_model.run(None, input_feeds) File "/home/anaconda3/envs/live-portrait/lib/python3.10/site-packages/onnxruntime/capi/onnxruntime_inference_collection.py", line 220, in run return self._sess.run(output_names, input_feed, run_options) onnxruntime.capi.onnxruntime_pybind11_state.InvalidArgument: [ONNXRuntimeError] : 2 : INVALID_ARGUMENT : Non-zero status code returned while running GridSample node. Name:'/dense_motion_network/GridSample' Status Message: Only 4-D tensor is supported

This is the traceback

[warmshao]

but i have had this error, do you know about it: onnxruntime.capi.onnxruntime_pybind11_state.InvalidArgument: [ONNXRuntimeError] : 2 : INVALID_ARGUMENT : Non-zero status code returned while running GridSample node. Name:'/dense_motion_network/GridSample' Status Message: Only 4-D tensor is supported

I suspect you haven't switched to the liqun/ImageDecoder-cuda branch.

[nldhuyen0047]

Hmmm, I'm sure that I did it. I switched to the liqun/ImageDecoder-cuda branch on github and downloaded it, so I could run by GPU but it had errors.

[warmshao]

Using onnxruntime-gpu as the backend, an alpha version for Windows has been made. You guys can start by helping to experience it. After extracting, just run app.bat directly, and let me know if there are any issues. Please note that it is assumed you have already installed CUDA and cuDNN, and have set up the environment variables properly.

https://drive.google.com/file/d/1ijqDlMAYqAVlqwqlXDpjBS5i3A6R_f7M/view?usp=sharing

[warmshao]

I have verified that transferring the Python virtual environment to another Windows computer does indeed cause some issues. I am still working on resolving this. Thank you for your feedback.

[nldhuyen0047]

I created a virtual environment and installed packages. I followed what you instructed at "ONNX Inference" to run by GPU on Linu

I created a virtual environment and installed packages. I followed what you instructed at "ONNX Inference" to run by GPU on Linux

How about using Docker? It's much simpler that way

Because I want to run on my local, I do not try running by Docker. Hmmm, but I think I will try it

[nldhuyen0047]

I tested it on linux operating system, if you have any solutions, please let me know. Thank you so muchhhhh

[nldhuyen0047]

@warmshao I think it's better because I use onnxruntime gpu version 1.7 instead of version 1.8 but it has an error such as:

Traceback (most recent call last): File "/home/code/FasterLivePortrait/run.py", line 65, in dri_crop = cv2.resize(dri_crop, (512, 512)) cv2.error: OpenCV(4.10.0) /io/opencv/modules/imgproc/src/resize.cpp:4152: error: (-215:Assertion failed) !ssize.empty() in function 'resize'

Everything was ok, from upload image, I choose dri_video 0 and runtime

[nldhuyen0047]

Hi @warmshao the issue is that I use onnxruntime version 18 instead of onnxruntime version 17

[warmshao]

Hi @warmshao the issue is that I use onnxruntime version 18 instead of onnxruntime version 17

can you run successfully？

[warmshao]

Hi guys, Install-free, extract-and-play Windows package with TensorRT support now available! Please watch change log, really fast!!!

[nldhuyen0047]

Hi @warmshao the issue is that I use onnxruntime version 18 instead of onnxruntime version 17

can you run successfully？

yah, I can run successfully

[warmshao]

enjoy～