【AI实战】YOLOv7加载多GPU训练的模型报错解决方法

最新推荐文章于 2024-07-01 17:26:37 发布
最新推荐文章于 2024-07-01 17:26:37 发布
阅读量1.7k 收藏
点赞数 1
分类专栏： 深度学习 人工智能 图像识别 文章标签： 深度学习 yolov7 多GPU
版权声明：本文为博主原创文章，遵循 CC 4.0 BY-SA 版权协议，转载请附上原文出处链接和本声明。
本文链接：https://blog.csdn.net/zengNLP/article/details/128626149
版权

【AI实战】YOLOv7加载多GPU训练的模型DataParallel

问题描述

使用多gpu训练YOLOv7完成后，加载模型进行单张图片推理时，报错如下：

Traceback (most recent call last):
  File "demo_detect.py", line 242, in <module>
    model =  DetectModel(api_path)
  File "demo_detect.py", line 50, in __init__
    self.model = attempt_load(self.weights, map_location=self.device)  # load FP32 model
  File "/workspace/code/2train_test_infer/yolov7/yolo7_fire_deploy/models/experimental.py", line 245, in attempt_load
    model.append(ckpt['ema' if ckpt.get('ema') else 'model'].float().fuse().eval())  # FP32 model
  File "/opt/conda/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1131, in __getattr__
    raise AttributeError("'{}' object has no attribute '{}'".format(
AttributeError: 'DataParallel' object has no attribute 'get'

打印网络信息：

DataParallel(
  (module): Model(
    (model): Sequential(
      (0): Conv(
        (conv): Conv2d(3, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(32, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (1): Conv(
        (conv): Conv2d(32, 64, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(64, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (2): Conv(
        (conv): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(64, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (3): Conv(
        (conv): Conv2d(64, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(128, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (4): Conv(
        (conv): Conv2d(128, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn): BatchNorm2d(64, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (5): Conv(
        (conv): Conv2d(128, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn): BatchNorm2d(64, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (6): Conv(
        (conv): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(64, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (7): Conv(
        (conv): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(64, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (8): Conv(
        (conv): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(64, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (9): Conv(
        (conv): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(64, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (10): Concat()
      (11): Conv(
        (conv): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn): BatchNorm2d(256, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (12): MP(
        (m): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
      )
      (13): Conv(
        (conv): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn): BatchNorm2d(128, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (14): Conv(
        (conv): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn): BatchNorm2d(128, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (15): Conv(
        (conv): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(128, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (16): Concat()
      (17): Conv(
        (conv): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn): BatchNorm2d(128, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (18): Conv(
        (conv): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn): BatchNorm2d(128, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (19): Conv(
        (conv): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(128, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (20): Conv(
        (conv): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(128, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (21): Conv(
        (conv): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(128, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (22): Conv(
        (conv): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(128, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (23): Concat()
      (24): Conv(
        (conv): Conv2d(512, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn): BatchNorm2d(512, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (25): MP(
        (m): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
      )
      (26): Conv(
        (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn): BatchNorm2d(256, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (27): Conv(
        (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn): BatchNorm2d(256, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (28): Conv(
        (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(256, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (29): Concat()
      (30): Conv(
        (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn): BatchNorm2d(256, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (31): Conv(
        (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn): BatchNorm2d(256, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (32): Conv(
        (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(256, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (33): Conv(
        (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(256, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (34): Conv(
        (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(256, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (35): Conv(
        (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(256, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (36): Concat()
      (37): Conv(
        (conv): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn): BatchNorm2d(1024, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (38): MP(
        (m): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
      )
      (39): Conv(
        (conv): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn): BatchNorm2d(512, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (40): Conv(
        (conv): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn): BatchNorm2d(512, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (41): Conv(
        (conv): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(512, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (42): Concat()
      (43): Conv(
        (conv): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn): BatchNorm2d(256, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (44): Conv(
        (conv): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn): BatchNorm2d(256, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (45): Conv(
        (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(256, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (46): Conv(
        (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(256, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (47): Conv(
        (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(256, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (48): Conv(
        (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(256, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (49): Concat()
      (50): Conv(
        (conv): Conv2d(1024, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn): BatchNorm2d(1024, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (51): SPPCSPC(
        (cv1): Conv(
          (conv): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn): BatchNorm2d(512, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
          (act): SiLU()
        )
        (cv2): Conv(
          (conv): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn): BatchNorm2d(512, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
          (act): SiLU()
        )
        (cv3): Conv(
          (conv): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn): BatchNorm2d(512, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
          (act): SiLU()
        )
        (cv4): Conv(
          (conv): Conv2d(512, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn): BatchNorm2d(512, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
          (act): SiLU()
        )
        (m): ModuleList(
          (0): MaxPool2d(kernel_size=5, stride=1, padding=2, dilation=1, ceil_mode=False)
          (1): MaxPool2d(kernel_size=9, stride=1, padding=4, dilation=1, ceil_mode=False)
          (2): MaxPool2d(kernel_size=13, stride=1, padding=6, dilation=1, ceil_mode=False)
        )
        (cv5): Conv(
          (conv): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn): BatchNorm2d(512, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
          (act): SiLU()
        )
        (cv6): Conv(
          (conv): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn): BatchNorm2d(512, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
          (act): SiLU()
        )
        (cv7): Conv(
          (conv): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn): BatchNorm2d(512, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
          (act): SiLU()
        )
      )
      (52): Conv(
        (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn): BatchNorm2d(256, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (53): Upsample(scale_factor=2.0, mode=nearest)
      (54): Conv(
        (conv): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn): BatchNorm2d(256, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (55): Concat()
      (56): Conv(
        (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn): BatchNorm2d(256, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (57): Conv(
        (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn): BatchNorm2d(256, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (58): Conv(
        (conv): Conv2d(256, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(128, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (59): Conv(
        (conv): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(128, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (60): Conv(
        (conv): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(128, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (61): Conv(
        (conv): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(128, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (62): Concat()
      (63): Conv(
        (conv): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn): BatchNorm2d(256, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (64): Conv(
        (conv): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn): BatchNorm2d(128, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (65): Upsample(scale_factor=2.0, mode=nearest)
      (66): Conv(
        (conv): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn): BatchNorm2d(128, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (67): Concat()
      (68): Conv(
        (conv): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn): BatchNorm2d(128, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (69): Conv(
        (conv): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn): BatchNorm2d(128, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (70): Conv(
        (conv): Conv2d(128, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(64, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (71): Conv(
        (conv): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(64, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (72): Conv(
        (conv): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(64, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (73): Conv(
        (conv): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(64, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (74): Concat()
      (75): Conv(
        (conv): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn): BatchNorm2d(128, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (76): MP(
        (m): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
      )
      (77): Conv(
        (conv): Conv2d(128, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn): BatchNorm2d(128, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (78): Conv(
        (conv): Conv2d(128, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn): BatchNorm2d(128, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (79): Conv(
        (conv): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(128, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (80): Concat()
      (81): Conv(
        (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn): BatchNorm2d(256, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (82): Conv(
        (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn): BatchNorm2d(256, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (83): Conv(
        (conv): Conv2d(256, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(128, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (84): Conv(
        (conv): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(128, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (85): Conv(
        (conv): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(128, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (86): Conv(
        (conv): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(128, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (87): Concat()
      (88): Conv(
        (conv): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn): BatchNorm2d(256, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (89): MP(
        (m): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
      )
      (90): Conv(
        (conv): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn): BatchNorm2d(256, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (91): Conv(
        (conv): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn): BatchNorm2d(256, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (92): Conv(
        (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(256, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (93): Concat()
      (94): Conv(
        (conv): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn): BatchNorm2d(512, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (95): Conv(
        (conv): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn): BatchNorm2d(512, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (96): Conv(
        (conv): Conv2d(512, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(256, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (97): Conv(
        (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(256, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (98): Conv(
        (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(256, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (99): Conv(
        (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(256, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (100): Concat()
      (101): Conv(
        (conv): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (bn): BatchNorm2d(512, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (102): RepConv(
        (act): SiLU()
        (rbr_dense): Sequential(
          (0): Conv2d(128, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (1): BatchNorm2d(256, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        )
        (rbr_1x1): Sequential(
          (0): Conv2d(128, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (1): BatchNorm2d(256, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        )
      )
      (103): RepConv(
        (act): SiLU()
        (rbr_dense): Sequential(
          (0): Conv2d(256, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (1): BatchNorm2d(512, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        )
        (rbr_1x1): Sequential(
          (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (1): BatchNorm2d(512, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        )
      )
      (104): RepConv(
        (act): SiLU()
        (rbr_dense): Sequential(
          (0): Conv2d(512, 1024, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (1): BatchNorm2d(1024, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        )
        (rbr_1x1): Sequential(
          (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (1): BatchNorm2d(1024, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        )
      )
      (105): IDetect(
        (m): ModuleList(
          (0): Conv2d(256, 30, kernel_size=(1, 1), stride=(1, 1))
          (1): Conv2d(512, 30, kernel_size=(1, 1), stride=(1, 1))
          (2): Conv2d(1024, 30, kernel_size=(1, 1), stride=(1, 1))
        )
        (ia): ModuleList(
          (0): ImplicitA()
          (1): ImplicitA()
          (2): ImplicitA()
        )
        (im): ModuleList(
          (0): ImplicitM()
          (1): ImplicitM()
          (2): ImplicitM()
        )
      )
    )
  )
)

解决方法

  • 修改：models/experimental.py
def attempt_load(weights, map_location=None):
    # Loads an ensemble of models weights=[a,b,c] or a single model weights=[a] or weights=a
    model = Ensemble()
    for w in weights if isinstance(weights, list) else [weights]:
        attempt_download(w)
        ckpt = torch.load(w, map_location=map_location)  # load
        model.append(ckpt['ema' if ckpt.get('ema') else 'model'].float().fuse().eval())  # FP32 model
        

    # Compatibility updates
    for m in model.modules():
        if type(m) in [nn.Hardswish, nn.LeakyReLU, nn.ReLU, nn.ReLU6, nn.SiLU]:
            m.inplace = True  # pytorch 1.7.0 compatibility
        elif type(m) is nn.Upsample:
            m.recompute_scale_factor = None  # torch 1.11.0 compatibility
        elif type(m) is Conv:
            m._non_persistent_buffers_set = set()  # pytorch 1.6.0 compatibility

    if len(model) == 1:
        return model[-1]  # return model
    else:
        print('Ensemble created with %s\n' % weights)
        for k in ['names', 'stride']:
            setattr(model, k, getattr(model[-1], k))
        return model  # return ensemble

修改为：

def attempt_load(weights, map_location=None):
    # Loads an ensemble of models weights=[a,b,c] or a single model weights=[a] or weights=a
    model = Ensemble()
    for w in weights if isinstance(weights, list) else [weights]:
        attempt_download(w)
        ckpt = torch.load(w, map_location=map_location)  # load
        print('ckpt', ckpt)
        try:
            model.append(ckpt['ema' if ckpt.get('ema') else 'model'].float().fuse().eval())  # FP32 model
        except:
            print(type(ckpt.module))
            model.append(ckpt.module.float().fuse().eval())  # FP32 model

    # Compatibility updates
    for m in model.modules():
        if type(m) in [nn.Hardswish, nn.LeakyReLU, nn.ReLU, nn.ReLU6, nn.SiLU]:
            m.inplace = True  # pytorch 1.7.0 compatibility
        elif type(m) is nn.Upsample:
            m.recompute_scale_factor = None  # torch 1.11.0 compatibility
        elif type(m) is Conv:
            m._non_persistent_buffers_set = set()  # pytorch 1.6.0 compatibility

    if len(model) == 1:
        return model[-1]  # return model
    else:
        print('Ensemble created with %s\n' % weights)
        for k in ['names', 'stride']:
            setattr(model, k, getattr(model[-1], k))
        return model  # return ensemble
  • 修改：models/yolo.py
    def fuse(self):
        print("IDetect.fuse")
        # fuse ImplicitA and Convolution
        for i in range(len(self.m)):
            c1,c2,_,_ = self.m[i].weight.shape
            c1_,c2_, _,_ = self.ia[i].implicit.shape
            self.m[i].bias += torch.matmul(self.m[i].weight.reshape(c1,c2),self.ia[i].implicit.reshape(c2_,c1_)).squeeze(1)

        # fuse ImplicitM and Convolution
        for i in range(len(self.m)):
            c1,c2, _,_ = self.im[i].implicit.shape
            self.m[i].bias *= self.im[i].implicit.reshape(c2)
                self.m[i].weight *= self.im[i].implicit.transpose(0,1)

修改为：

    def fuse(self):
        print("IDetect.fuse")
        # fuse ImplicitA and Convolution
        for i in range(len(self.m)):
            with torch.no_grad():
                c1,c2,_,_ = self.m[i].weight.shape
                c1_,c2_, _,_ = self.ia[i].implicit.shape
                self.m[i].bias += torch.matmul(self.m[i].weight.reshape(c1,c2),self.ia[i].implicit.reshape(c2_,c1_)).squeeze(1)

        # fuse ImplicitM and Convolution
        for i in range(len(self.m)):
            with torch.no_grad():
                c1,c2, _,_ = self.im[i].implicit.shape
                self.m[i].bias *= self.im[i].implicit.reshape(c2)
                self.m[i].weight *= self.im[i].implicit.transpose(0,1)

再运行测试代码即可正常运行

szZack
关注
  • 1
    点赞
  • 踩
  • 0
    收藏
    觉得还不错? 一键收藏
  • 打赏
    打赏
  • 1
    评论
专栏目录
YOLO8训练遇到的各种报错及解决方案
专注于图像领域，主要研究内容包括计算机视觉和深度学习，特别是在图像分类、目标检测和图像生成等方面有深入的研究和实践经验。
08-06 127
在Windows系统中，由于其多进程模型的特殊性，使用多进程加载数据时经常会遇到问题。在Linux系统中，通常可以使用多个子进程进行数据加载，但在Windows系统中必须将进程数设置为单进程来避免错误。此问题可能是当前使用的GPU正在被其他应用程序占用，导致显存不足。需要关闭其他应用程序以释放显存。这是由于Pillow版本问题，最新的10.0版本会产生上述问题。虽然调整后会在一定程度上影响模型的泛化能力并容易导致过拟合，但在必要时可以将。降低PyTorch和CUDA版本。使用device指定GPU。
《YOLOv8：从入门到实战》报错解决 & 专栏答疑
突然好想你
02-27 668
《YOLOv8：从入门到实战》专栏上线后，部分同学在学习过程中提出了一些问题，笔者相信这些问题其他同学也有可能遇到。为了让大家可以更好地学习本专栏内容，笔者特意推出了该篇专栏答疑，针对同学们在学习过程中所提出的问题进行汇总记录，并不断实时更新，希望能够帮助到大家！🌈
Python Apex YOLO V7 main 目标检测 全过程记录
mrathena
10-29 4297
我直接使用了 YOLO V5 6.2 的环境, 什么都不需要动, 直接 GPU 推测. 可参考上文 环境准备 部分。找到报错文件, 找到报错行, 修改为。期间有一个警告, 解决方法如下。即可, 该参数默认值就是。
[YOLOv8单机多卡GPU问题解决]
dujuancao11的博客
06-26 451
解决：MKL_THREADING_LAYER=GPU。分析：直接定位到Error部分。
【解决方案】yolov8&yolov10多卡训练报错
最新发布
VAN的博客
07-01 1161
【解决方案】yolov8&yolov10多卡训练报错
labview yolov5 onnxruntime推理，封装dll, labview调用dll，支持同时加载多个模型并行推理
xHttkHeu的博客
10-23 588
LabVIEW YOLOv5 ONNXRuntime推理，封装DLL，LabVIEW调用DLL，支持同时加载多个模型并行推理，可CPU GPU，x86 x64位，识别视频和图片，CPU速度100ms，GPU26ms，只需要替换模型的ONNX和nameclass即可。通过本文的学习，读者将掌握使用LabVIEW进行物体识别的方法和技巧，并能够根据需求进行调整和优化，实现更高效、更准确的物体探测。最后，还将阐述如何在不同硬件环境下进行模型的部署以及如何替换模型的ONNX和nameclass文件。
YOLOv7环境安装及训练时的一些报错及解决方法
Tiger_DAHE的博客
04-08 2186
作者在实验的时候遇到了很多困难，写在这里希望能帮到更多人
YOLOv7使用云GPU训练自己的数据集
weixin_43764678的博客
07-19 6128
YOLOv7使用云GPU训练自己的数据集
关于yolo7和gpu
whaosoft143ai的博客
08-03 1895
OneEmbedding 是应训练大规模推荐系统模型需求而生的组件，灵活的分层存储、高度优化的数据流水线以及易于横向扩展的特性，能让用户轻松训练 TB 级别的推荐模型。目前，OneFlow 框架提供了一些模型示例供你一键体验 OneEmbedding。后续，OneFlow 团队将推出覆盖业界主流模型的推荐系统模型库 Flow-Recommender ，它不仅支持分布式训练，还支持分布式推理。欢迎感兴趣的朋友关注。...
YoloAll V2发布，集成所有主流Yolo模型于一身
cjnewstar111的专栏
11-26 3024
YoloAll项目简介 YoloAll是一个将当前主流Yolo版本集成到同一个UI界面下的推理预测工具。可以迅速切换不同的yolo版本，并且可以针对图片，视频，摄像头码流进行实时推理，可以很方便，直观的对比不同版本的yolo的推理效果，耗时等。 自从YoloAll V1版本推出以后，得到了很多朋友的喜欢，大家也跟我一样，被众多的Yolo版本所困扰，有时为了比较两个不同版本的Yolo的效果，往往需要花费很长时间搭建环境，并且很容易出错，因此才有了这个开发这个YoloAll的想法，能够非常方便的测试不同..
手把手教你搭建YOLOV8+CUDA环境，训练自定义数据集，训练推理验证导出。小白也能看得懂的！
专注于图像领域，主要研究内容包括计算机视觉和深度学习，特别是在图像分类、目标检测和图像生成等方面有深入的研究和实践经验。
05-18 5164
YOLO（You Only Look Once）系列算法因其高效、准确等特点而备受瞩目。由2023年Ultralytics公司发布了YOLO的最新版本YOLOv8是结合前几代YOLO的基础上的一个融合改进版。本文主要介绍YOLOv8环境搭建/数据集获取/训练/推理/验证/导出/部署
YOLOv5项目实战（1）— 如何去训练模型
突然好想你
10-23 540
为了让大家能够牢固地掌握YOLOv5算法，本系列文章就通过一个个精彩案例来讲解YOLOv5在工作或日常生活中的应用。在开始讲解之前，让我们先大概预习一下如何去训练模型，希望大家学习之后能够有所收获！
yolov7目标检测：基于自定义数据集完成检测、训练、测试
shinuone的博客
03-25 5460
主要分三步：（1）环境配置与文件配置（2）检测（3）训练。其中，检测和训练都是可以独立进行的。检测是依赖于权重文件即可运行，而训练是基于自定义训练数据集和超参数生成权重文件。
yolov5同时加载两个模型实现推理生成标签
weixin_45994963的博客
12-12 5309
yolov5同时加载两个模型实现推理生成标签
已解决训练自己yolov7检测模型时报错：AssertionError: train: No labels in XXX/XXX/train.cache.
ff_5_5_的博客
03-18 3452
但是经过上步修改后运行train.py还会报相同的错误时，这是因为前面运行了train.py文件，已在VOCdevkit/VOC2007文件夹下生成。（VOCdevkit/VOC2007是我存放自己数据集的路径）文件（不完整的），需要在下次运行前将train.cache文件。经过上面两个操作后重新运行train.py网络就开始训练了。函数中默认存放数据集图像的文件夹是images，需要。进行调试后，发现yolov7源代码在。，就能根据图像位置获得标签位置。
yolov7训练自己的数据集-gpu版
qq_62975494的博客
03-27 5029
总共有两个文件需要配置，一个是/yolov7/cfg/training/yolov7.yaml，这个文件是有关模型的配置文件；一个是/yolov7/data/coco.yaml，这个是数据集的配置文件。1.第一步，复制yolov7.yaml文件到相同的路径下，然后重命名，我们重命名为yolov7-xxx.yaml(起一个你想要的名字)然后打开。2.复制coco.yaml文件到相同的路径下，然后重命名，我们命名为xxx.yaml。标注完成后如图所示labels中的txt文件对应images中的图片。
YOLOV7 DP可以，但是DDP老是报错怎么办？
Mrwei_418的博客
05-05 424
解决yolov7 DDP报错！！！
pytorch加载10个yolov5模型并推理1万图片
weixin_38132142的博客
03-25 575
【代码】pytorch加载10个yolov5模型并推理1万图片。
Multi-GPU Training多GPU训练（Yolo系列）
fankeyang的博客
11-17 764
通过简单地传递后跟您的特定 GPU 来做到这一点。例如，在下面的代码中，我们将使用 GPU .
yolov7预训练模型
08-25
yolov7预训练模型可以通过以下步骤进行获取和部署： 1. 首先，使用引用[2]中给出的命令来训练模型。这将使用指定的数据集、配置文件和超参数来进行训练，生成一个训练好的模型的权重文件。 2. 接下来，根据引用中提供的链接下载预训练模型的权重文件yolov7.pt。这个文件可以作为一个已经在大规模数据集上预训练好的模型。 3. 一旦下载完成，你可以使用这个预训练模型进行目标检测任务。执行引用中的命令来生成一个yolov7-tiny.onnx文件，这个文件将用于在TensorRT中进行部署。 4. 最后，使用TensorRT来部署模型。你可以按照TensorRT的文档和示例来加载和推理这个预训练模型。这将允许你在生产环境中使用这个模型来进行目标检测。 综上所述，yolov7预训练模型可以通过训练模型、下载预训练权重文件、生成ONNX文件和使用TensorRT进行部署的步骤来获取和使用。<span class="em">1</span><span class="em">2</span><span class="em">3</span> #### 引用[.reference_title] - *1* *2* [Yolov7模型训练与部署](https://blog.csdn.net/iamqianrenzhan/article/details/127178904)[target="_blank" data-report-click={"spm":"1018.2226.3001.9630","extra":{"utm_source":"vip_chatgpt_common_search_pc_result","utm_medium":"distribute.pc_search_result.none-task-cask-2~all~insert_cask~default-1-null.142^v93^chatsearchT3_1"}}] [.reference_item style="max-width: 50%"] - *3* [足球视频AI(三)——YOLOV7目标检测自训练模型](https://blog.csdn.net/black0707/article/details/128549772)[target="_blank" data-report-click={"spm":"1018.2226.3001.9630","extra":{"utm_source":"vip_chatgpt_common_search_pc_result","utm_medium":"distribute.pc_search_result.none-task-cask-2~all~insert_cask~default-1-null.142^v93^chatsearchT3_1"}}] [.reference_item style="max-width: 50%"] [ .reference_list ]
评论 1
添加红包

请填写红包祝福语或标题

个

红包个数最小为10个

元

红包金额最低5元

当前余额3.43元 前往充值 >
需支付：10.00元
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包

打赏作者

szZack

你的鼓励将是我创作的最大动力

¥1 ¥2 ¥4 ¥6 ¥10 ¥20
扫码支付：¥1
获取中
扫码支付

您的余额不足，请更换扫码支付或充值

打赏作者

实付元
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明：

1.余额是钱包充值的虚拟货币，按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载，可以购买VIP、付费专栏及课程。

余额充值