快30倍以上的切脸软件(cpu和N卡版本)

huaxin123

lei27028 发表于 2023-2-26 12:16
这种情况我也遇到了，gpu选1就可以了

咋选呢，没地方说要选啊

huaxin123

这个bug啥意思呢，看不懂啊

huaxin123

huaxin123 发表于 2023-7-5 20:38
这个bug啥意思呢，看不懂啊

搞定了，试了好几个cuda版本，然后看留言用了11.8就ok了

fanda

huaxin123 发表于 2023-7-5 21:30
搞定了，试了好几个cuda版本，然后看留言用了11.8就ok了

版本的问题，这个是玄学

1315

不能把这些CUDA什么的打包集合在一起吗？解压即用不好吗

laiking888

大大用这个软件,没有aligned_debug文件

wuyan

切脸中途出现错误：
人脸提取进度: 1301it [00:44, 24.85it/s]D:\tools\DFL_maozhihanhua_RTX2080Ti\dfl_code\insightface\model_zoo\landmark.py:85: RuntimeWarning: divide by zero encountered in scalar divide
  _scale = self.input_size[0]  / (max(w, h)*1.5)
D:\tools\DFL_maozhihanhua_RTX2080Ti\dfl_code\insightface\python_env\lib\site-packages\skimage\transform\_geometric.py:1333: RuntimeWarning: invalid value encountered in multiply
  matrix[:dimensionality, :dimensionality] *= scale
D:\tools\DFL_maozhihanhua_RTX2080Ti\dfl_code\insightface\python_env\lib\site-packages\skimage\transform\_geometric.py:765: RuntimeWarning: invalid value encountered in matmul
  return tform(other.params @ self.params)
使用GPU切脸，线程5

人脸提取进度: 1071it [00:16, 42.15it/s]2023-07-07 17:36:28.7463739 [Ennxruntimeefault, cuda_call.cc:119 onnxruntime::CudaCall] CUDA failure 700: an illegal memory access was encountered ; GPU=0 ; hostname=WU ; expr=cudaMemcpyAsync(dst_data, src_data, bytes, cudaMemcpyHostToDevice, static_cast<cudaStream_t>(stream.GetHandle()));
2023-07-07 17:36:28.7494662 [Ennxruntimeefault, cuda_call.cc:119 onnxruntime::CudaCall] CUDA failure 700: an illegal memory access was encountered ; GPU=0 ; hostname=WU ; expr=cudaStreamSynchronize(static_cast<cudaStream_t>(stream_));
Traceback (most recent call last):
  File "frames-extract-face.py", line 243, in face_det_lmk_dir
    future.result()
  File "D:\tools\DFL_maozhihanhua_RTX2080Ti\dfl_code\insightface\python_env\lib\concurrent\futures\_base.py", line 437, in result
    return self.__get_result()
  File "D:\tools\DFL_maozhihanhua_RTX2080Ti\dfl_code\insightface\python_env\lib\concurrent\futures\_base.py", line 389, in __get_result
    raise self._exception
  File "D:\tools\DFL_maozhihanhua_RTX2080Ti\dfl_code\insightface\python_env\lib\concurrent\futures\thread.py", line 57, in run
    result = self.fn(*self.args, **self.kwargs)
  File "frames-extract-face.py", line 153, in extract_to_dfl_img
    faces = self.faceAnalysis.get(img_mat)
  File "D:\tools\DFL_maozhihanhua_RTX2080Ti\dfl_code\insightface\app\face_analysis.py", line 62, in get
    bboxes, kpss = self.det_model.detect(img,
  File "D:\tools\DFL_maozhihanhua_RTX2080Ti\dfl_code\insightface\model_zoo\retinaface.py", line 226, in detect
    scores_list, bboxes_list, kpss_list = self.forward(det_img, self.det_thresh)
  File "D:\tools\DFL_maozhihanhua_RTX2080Ti\dfl_code\insightface\model_zoo\retinaface.py", line 154, in forward
    net_outs = self.session.run(self.output_names, {self.input_name : blob})
  File "D:\tools\DFL_maozhihanhua_RTX2080Ti\dfl_code\insightface\python_env\lib\site-packages\onnxruntime\capi\onnxruntime_inference_collection.py", line 200, in run
    return self._sess.run(output_names, input_feed, run_options)
onnxruntime.capi.onnxruntime_pybind11_state.Fail: [ONNXRuntimeError] : 1 : FAIL : CUDA failure 700: an illegal memory access was encountered ; GPU=0 ; hostname=WU ; expr=cudaMemcpyAsync(dst_data, src_data, bytes, cudaMemcpyHostToDevice, static_cast<cudaStream_t>(stream.GetHandle()));
2023-07-07 17:36:28.7919616 [Ennxruntimeefault, cuda_call.cc:119 onnxruntime::CudaCall] CUDA failure 700: an illegal memory access was encountered ; GPU=0 ; hostname=WU ; expr=cudaMemcpyAsync(dst_data, src_data, bytes, cudaMemcpyHostToDevice, static_cast<cudaStream_t>(stream.GetHandle()));

fanda

laiking888 发表于 2023-7-7 01:22
大大用这个软件,没有aligned_debug文件

有选yes么

fanda

wuyan 发表于 2023-7-7 17:35
切脸中途出现错误：
人脸提取进度: 1301it [00:44, 24.85it/s]D:\tools\DFL_maozhihanhua_RTX2080Ti\dfl_co ...

线程默认就行，省不了多少时间。。

laiking888

fanda 发表于 2023-7-7 20:10
有选yes么

明白,谢谢大大