search

dinglufe / segment-anything-cpp-wrapper

MIT License
221 stars 34 forks source link

模型问题，想问一下那个processing的作用是啥 #20

Closed lixinye666 closed 11 months ago

lixinye666 commented 1 year ago

sam_preprocess.onnx

lixinye666 commented 1 year ago

为啥用俩模型

dinglufe commented 1 year ago

The preprocessing here is approximately the embedding in SAM.

Segment Anything involves several preprocessing steps, like this:

sam.to(device='cuda')
predictor = SamPredictor(sam)
predictor.set_image(image)
image_embedding = predictor.get_image_embedding().cpu().numpy()

The export_pre_model script exports these operations as an ONNX model to enable execution independent of the Python environment.

lixinye666 commented 1 year ago

Progress: 0% 2023-07-26 17:04:50.4252918 [E:onnxruntime:test, cuda_call.cc:119 onnxruntime::CudaCall] CUDNN failure 4: CUDNN_STATUS_INTERNALERROR ; GPU=0 ; hostname=MSI ; expr=cudnnFindConvolutionForwardAlgorithmEx( GetCudnnHandle(context), s.xtensor, s.xdata, s.wdesc, s.wdata, s.convdesc, s.ytensor, s.y_data, 1, &algo_count, &perf, algo_search_workspace.get(), max_ws_size); 2023-07-26 17:04:50.4346841 [E:onnxruntime:, sequential_executor.cc:494 onnxruntime::ExecuteKernel] Non-zero status code returned while running Conv node. Name:'/mask_downscaling/mask_downscaling.0/Conv' Status Message: CUDNN failure 4: CUDNN_STATUS_INTERNALERROR ; GPU=0 ; hostname=MSI ; expr=cudnnFindConvolutionForwardAlgorithmEx( GetCudnnHandle(context), s.xtensor, s.xdata, s.wdesc, s.wdata, s.convdesc, s.ytensor, s.y_data, 1, &algo_count, &perf, algo_search_workspace.get(), max_ws_size);