无法使用NPU

[1556900941lizerui]

问题描述 / Problem Description

使用pytorch框架想要使用NPU时发生报错

运行环境 / Runtime Environment

CANN 8.0.0
rapidocr 3.4.2
torch 2.1
torch_npu 2.1

复现代码 / Reproduction Code

from rapidocr import EngineType, RapidOCR

import time 

#使用pytorch推理引擎
engine = RapidOCR(
    params={
        "Det.engine_type": EngineType.TORCH,
        "Cls.engine_type": EngineType.TORCH,
        "Rec.engine_type": EngineType.TORCH,
        "EngineConfig.torch.use_npu": True,  # 使用torch NPU版推理
        "EngineConfig.torch.npu_id": 0,  # 指定NPU id
    }
)

#使用onnx推理引擎
# engine = RapidOCR()

img_path = "/root/autodl-tmp/RapidOCR-main/assets/RapidOCR_LOGO.png"
start = time.time()
print("start to engine the image")
for i in range(5):
    result = engine(img_path)
   
    print(result)

end = time.time()
print(f"spend time is {end - start}")
result.vis('vis_result.jpg')

Traceback (most recent call last):
  File "/root/autodl-tmp/RapidOCR-main/test.py", line 6, in <module>
    engine = RapidOCR(
  File "/root/miniconda3/envs/classify/lib/python3.10/site-packages/rapidocr/main.py", line 43, in __init__
    self._initialize(cfg)
  File "/root/miniconda3/envs/classify/lib/python3.10/site-packages/rapidocr/main.py", line 64, in _initialize
    self.text_det = TextDetector(cfg.Det)
  File "/root/miniconda3/envs/classify/lib/python3.10/site-packages/rapidocr/ch_ppocr_det/main.py", line 45, in __init__
    self.session = get_engine(cfg.engine_type)(cfg)
  File "/root/miniconda3/envs/classify/lib/python3.10/site-packages/rapidocr/inference_engine/torch.py", line 27, in __init__
    self._setup_device(cfg)
  File "/root/miniconda3/envs/classify/lib/python3.10/site-packages/rapidocr/inference_engine/torch.py", line 74, in _setup_device
    self.device, self.use_gpu, self.use_npu = self._resolve_device_config(cfg)
  File "/root/miniconda3/envs/classify/lib/python3.10/site-packages/rapidocr/inference_engine/torch.py", line 86, in _resolve_device_config
    return torch.device(f"npu:{cfg.engine_cfg.npu_id}"), False, True
RuntimeError: Expected one of cpu, cuda, ipu, xpu, mkldnn, opengl, opencl, ideep, hip, ve, fpga, ort, xla, lazy, vulkan, mps, meta, hpu, mtia, privateuseone device type at start of device string: npu

可能解决方案 / Possible solutions

[SWHL]

先确定一下torch_npu是否安装成功