Python 如何将自定义Pytorch模型转换为torchscript（pth到pt模型）？

我使用colab环境使用pytorch训练了一个定制模型。我成功地将经过培训的模型保存到名为model_final.pth的Google Drive。我想将model_final.pth转换为model_final.pt，以便可以在移动设备上使用。 我用来训练模型的代码如下：

cfg = get_cfg()
cfg.merge_from_file(model_zoo.get_config_file("COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_3x.yaml"))
cfg.DATASETS.TRAIN = ("mouse_train",)
cfg.DATASETS.TEST = ()
cfg.DATALOADER.NUM_WORKERS = 2
cfg.MODEL.WEIGHTS = model_zoo.get_checkpoint_url("COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_3x.yaml") 
cfg.SOLVER.IMS_PER_BATCH = 2
cfg.SOLVER.BASE_LR = 0.00025 
cfg.SOLVER.MAX_ITER = 1000   
cfg.SOLVER.STEPS = []        
cfg.MODEL.ROI_HEADS.BATCH_SIZE_PER_IMAGE = 512   
cfg.MODEL.ROI_HEADS.NUM_CLASSES = 1  
cfg.OUTPUT_DIR="drive/Detectron2/"

os.makedirs(cfg.OUTPUT_DIR, exist_ok=True)
trainer = DefaultTrainer(cfg) 
trainer.resume_or_load(resume=False)
trainer.train()

import torch
import torchvision

print("cfg.MODEL.WEIGHTS: ",cfg.MODEL.WEIGHTS)   ## RETURNS : cfg.MODEL.WEIGHTS:  drive/Detectron2/model_final.pth
model = build_model(cfg)
model.eval()
example = torch.rand(1, 3, 224, 224)
traced_script_module = torch.jit.trace(model, example)
traced_script_module.save("drive/Detectron2/model-final.pt")

我用来转换模型的代码如下

cfg = get_cfg()
cfg.merge_from_file(model_zoo.get_config_file("COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_3x.yaml"))
cfg.DATASETS.TRAIN = ("mouse_train",)
cfg.DATASETS.TEST = ()
cfg.DATALOADER.NUM_WORKERS = 2
cfg.MODEL.WEIGHTS = model_zoo.get_checkpoint_url("COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_3x.yaml") 
cfg.SOLVER.IMS_PER_BATCH = 2
cfg.SOLVER.BASE_LR = 0.00025 
cfg.SOLVER.MAX_ITER = 1000   
cfg.SOLVER.STEPS = []        
cfg.MODEL.ROI_HEADS.BATCH_SIZE_PER_IMAGE = 512   
cfg.MODEL.ROI_HEADS.NUM_CLASSES = 1  
cfg.OUTPUT_DIR="drive/Detectron2/"

os.makedirs(cfg.OUTPUT_DIR, exist_ok=True)
trainer = DefaultTrainer(cfg) 
trainer.resume_or_load(resume=False)
trainer.train()

import torch
import torchvision

print("cfg.MODEL.WEIGHTS: ",cfg.MODEL.WEIGHTS)   ## RETURNS : cfg.MODEL.WEIGHTS:  drive/Detectron2/model_final.pth
model = build_model(cfg)
model.eval()
example = torch.rand(1, 3, 224, 224)
traced_script_module = torch.jit.trace(model, example)
traced_script_module.save("drive/Detectron2/model-final.pt")

但我得到了这个错误

/usr/local/lib/python3.6/dist-packages/torch/tensor.py:593: RuntimeWarning: Iterating over a tensor might cause the trace to be incorrect. Passing a tensor of different shape won't change the number of iterations executed (and might lead to errors or silently give incorrect results).
  'incorrect results).', category=RuntimeWarning)
---------------------------------------------------------------------------
IndexError                                Traceback (most recent call last)
<ipython-input-17-8e544c0f39c8> in <module>()
      7 model.eval()
      8 example = torch.rand(1, 3, 224, 224)
----> 9 traced_script_module = torch.jit.trace(model, example)
     10 traced_script_module.save("drive/Detectron2/model_final.pt")

7 frames
/usr/local/lib/python3.6/dist-packages/detectron2/modeling/meta_arch/rcnn.py in <listcomp>(.0)
    219         Normalize, pad and batch the input images.
    220         """
--> 221         images = [x["image"].to(self.device) for x in batched_inputs]
    222         images = [(x - self.pixel_mean) / self.pixel_std for x in images]
    223         images = ImageList.from_tensors(images, self.backbone.size_divisibility)

IndexError: too many indices for tensor of dimension 3

/usr/local/lib/python3.6/dist-packages/torch/tensor.py:593:RuntimeWarning:在tensor上迭代可能会导致跟踪不正确。传递不同形状的张量不会改变执行的迭代次数（并且可能会导致错误或给出错误的结果）。
“结果不正确）。”，类别=运行时警告）
---------------------------------------------------------------------------
索引器回溯（最后一次最近调用）
在（）
7.模型评估（）
8示例=torch.rand（1,3,224,224）
---->9跟踪的脚本模块=torch.jit.trace（模型，示例）
10跟踪的脚本模块保存（“驱动器/Detectron2/model\u final.pt”）
7帧
/usr/local/lib/python3.6/dist-packages/detectron2/modeling/meta_arch/rcnn.py in（.0）
219对输入图像进行规格化、填充和批处理。
220         """
-->221图像=[x[“图像”].到（self.device），用于批处理_输入中的x]
222个图像=[（x-self.pixel_平均值）/self.pixel_标准表示图像中的x]
223 images=ImageList.from_张量（images，self.backbone.size_整除性）
索引器：维度3的张量的索引太多
默认情况下，

因此，您不能直接使用

torch.jit.trace

函数。但它们提供了一个名为的包装器，允许模型将张量或张量元组作为输入。您可以了解如何在模型中使用它

跟踪掩码RCNN模型的代码可能是（我没有尝试）：

可以找到更多关于detectron2部署和跟踪的信息