devbox
Cpp五条
这个屌丝很懒，什么也没留下！
关注作者
热门标签
热门文章
当前位置:   article > 正文

YOLOv8训练报错FileNotFoundError: train: No labels found in ...VisDrone2019-DET-train/labels.cache

作者：Cpp五条 | 2024-05-24 15:26:16

赞

踩

filenotfounderror: train: no labels found in

报错信息是什么？

我在使用YOLOv8，在VisDrone数据集上训练的时候，在使用train.,py进行  数据下载和训练 的时候，会遇到下面这个报错，导致无法正常训练。

train.py里面写的内容是下面这些

  1. from ultralytics import YOLO
  2. model = YOLO("yolov8n.yaml") 
  3. model.train(data="VisDrone.yaml", epochs=1,batch=1)

完整的报错信息如下

  1.   File "电脑路径/ultralytics/yolo/data/dataset.py", line 114, in get_labels
  2.     raise FileNotFoundError(f"{self.prefix}No labels found in {cache_path}, can not start training. {HELP_URL}")
  3. FileNotFoundError: train: No labels found in /root/m_coding/yolov3/datasets/VisDrone/datasets/VisDrone/VisDrone2019-DET-train/labels.cache, can not start training. See https://github.com/ultralytics/yolov5/wiki/Train-Custom-Data

更完整的报错信息

        先是说，尝试进行了多次访问，都无法访问到那个下载链接

  1. Dataset 'VisDrone.yaml' not found ⚠️, missing paths ['/root/m_coding/yolov3/datasets/VisDrone/datasets/VisDrone/VisDrone2019-DET-val/images']
  2. Downloading https://github.com/ultralytics/yolov5/releases/download/v1.0/VisDrone2019-DET-train.zip to /root/m_coding/yolov3/datasets/VisDrone/datasets/VisDrone/VisDrone2019-DET-train.zip...
  3. Downloading https://github.com/ultralytics/yolov5/releases/download/v1.0/VisDrone2019-DET-val.zip to /root/m_coding/yolov3/datasets/VisDrone/datasets/VisDrone/VisDrone2019-DET-val.zip...
  4. Downloading https://github.com/ultralytics/yolov5/releases/download/v1.0/VisDrone2019-DET-test-dev.zip to /root/m_coding/yolov3/datasets/VisDrone/datasets/VisDrone/VisDrone2019-DET-test-dev.zip...
  5. Downloading https://github.com/ultralytics/yolov5/releases/download/v1.0/VisDrone2019-DET-test-challenge.zip to /root/m_coding/yolov3/datasets/VisDrone/datasets/VisDrone/VisDrone2019-DET-test-challenge.zip...
  6. curl: (56) Unexpected EOF
  7. ⚠️ Download failure, retrying 1/3 https://github.com/ultralytics/yolov5/releases/download/v1.0/VisDrone2019-DET-val.zip...
  8. curl: (35) OpenSSL SSL_connect: SSL_ERROR_SYSCALL in connection to github.com:443 
  9. ⚠️ Download failure, retrying 2/3 https://github.com/ultralytics/yolov5/releases/download/v1.0/VisDrone2019-DET-val.zip...
  10. curl: (56) Unexpected EOF
  11. ⚠️ Download failure, retrying 1/3 https://github.com/ultralytics/yolov5/releases/download/v1.0/VisDrone2019-DET-train.zip...
  12. Unzipping /root/m_coding/yolov3/datasets/VisDrone/datasets/VisDrone/VisDrone2019-DET-val.zip to /root/m_coding/yolov3/datasets/VisDrone/datasets/VisDrone...
  13. curl: (56) Unexpected EOF
  14. ⚠️ Download failure, retrying 1/3 https://github.com/ultralytics/yolov5/releases/download/v1.0/VisDrone2019-DET-test-challenge.zip...
  15. curl: (56) Unexpected EOF
  16. ⚠️ Download failure, retrying 2/3 https://github.com/ultralytics/yolov5/releases/download/v1.0/VisDrone2019-DET-train.zip...
  17. curl: (56) Unexpected EOF
  18. ⚠️ Download failure, retrying 1/3 https://github.com/ultralytics/yolov5/releases/download/v1.0/VisDrone2019-DET-test-dev.zip...
  19. curl: (7) Failed to connect to github.com port 443: Connection timed out
  20. curl: (35) OpenSSL SSL_connect: SSL_ERROR_SYSCALL in connection to objects.githubusercontent.com:443 
  21. ⚠️ Download failure, retrying 2/3 https://github.com/ultralytics/yolov5/releases/download/v1.0/VisDrone2019-DET-test-dev.zip...
  22. curl: (56) Unexpected EOF
  23. ⚠️ Download failure, retrying 3/3 https://github.com/ultralytics/yolov5/releases/download/v1.0/VisDrone2019-DET-train.zip...
  24. Unzipping /root/m_coding/yolov3/datasets/VisDrone/datasets/VisDrone/VisDrone2019-DET-test-challenge.zip to /root/m_coding/yolov3/datasets/VisDrone/datasets/VisDrone...
  25. curl: (56) Unexpected EOF
  26. ⚠️ Download failure, retrying 3/3 https://github.com/ultralytics/yolov5/releases/download/v1.0/VisDrone2019-DET-test-dev.zip...
  27. curl: (56) Unexpected EOF

        然后说找不到labels.cache 标签文件

  1. train: Scanning /root/m_coding/yolov3/datasets/VisDrone/datasets/VisDrone/VisDrone2019-DET-train/labels... 0 images, 6471 backgrounds, 0 corrup
  2. train: WARNING ⚠️ No labels found in /root/m_coding/yolov3/datasets/VisDrone/datasets/VisDrone/VisDrone2019-DET-train/labels.cache. See https://github.com/ultralytics/yolov5/wiki/Train-Custom-Data
  3. train: New cache created: /root/m_coding/yolov3/datasets/VisDrone/datasets/VisDrone/VisDrone2019-DET-train/labels.cache
  4. Traceback (most recent call last):
  5.   File "train.py", line 3, in <module>
  6.     model.train(data="VisDrone.yaml", epochs=1,batch=1)  
  7.   File "/root/m_coding/YOLOv8n+LADHead/ultralytics/yolo/engine/model.py", line 251, in train
  8.     self.trainer.train()
  9.   File "/root/m_coding/YOLOv8n+LADHead/ultralytics/yolo/engine/trainer.py", line 187, in train
  10.     self._do_train(RANK, world_size)
  11.   File "/root/m_coding/YOLOv8n+LADHead/ultralytics/yolo/engine/trainer.py", line 252, in _do_train
  12.     self._setup_train(rank, world_size)
  13.   File "/root/m_coding/YOLOv8n+LADHead/ultralytics/yolo/engine/trainer.py", line 237, in _setup_train
  14.     self.train_loader = self.get_dataloader(self.trainset, batch_size=batch_size, rank=rank, mode="train")
  15.   File "/root/m_coding/YOLOv8n+LADHead/ultralytics/yolo/v8/detect/train.py", line 42, in get_dataloader
  16.     build_dataloader(self.args, batch_size, img_path=dataset_path, stride=gs, rank=rank, mode=mode,
  17.   File "/root/m_coding/YOLOv8n+LADHead/ultralytics/yolo/data/build.py", line 71, in build_dataloader
  18.     dataset = YOLODataset(
  19.   File "/root/m_coding/YOLOv8n+LADHead/ultralytics/yolo/data/dataset.py", line 44, in __init__
  20.     super().__init__(img_path, imgsz, cache, augment, hyp, prefix, rect, batch_size, stride, pad, single_cls)
  21.   File "/root/m_coding/YOLOv8n+LADHead/ultralytics/yolo/data/base.py", line 49, in __init__
  22.     self.labels = self.get_labels()
  23.   File "/root/m_coding/YOLOv8n+LADHead/ultralytics/yolo/data/dataset.py", line 114, in get_labels
  24.     raise FileNotFoundError(f"{self.prefix}No labels found in {cache_path}, can not start training. {HELP_URL}")
  25. FileNotFoundError: train: No labels found in /root/m_coding/yolov3/datasets/VisDrone/datasets/VisDrone/VisDrone2019-DET-train/labels.cache, can not start training. See https://github.com/ultralytics/yolov5/wiki/Train-Custom-Data

解释报错的原因

出现上面这种报错的情况下，你打开 训练集 验证集 测试集的文件夹会发现下面这种情况。只有image图片和annnotations标注，但是没有labels这个文件夹

正常情况下应该是下面这样，训练集、验证集、测试集 都是有 ultralytics加工过的labels文件夹和label.cache这个缓存文件。

labels这个文件夹,并不是天津大学VisDrone官方提供的。这个labels文件夹是你在运行上面train.py的这句话“model.train(data="VisDrone.yaml", epochs=1,batch=1)   ”，执行了"VisDrone.yaml"里面 visdrone2yolo这个函数  下面这些加工处理annotations文件夹中的标注数据后得到的

  1.   def visdrone2yolo(dir):
  2.       from PIL import Image
  3.       from tqdm import tqdm
  4.  
  5.       def convert_box(size, box):
  6.           # Convert VisDrone box to YOLO xywh box
  7.           dw = 1. / size[0]
  8.           dh = 1. / size[1]
  9.           return (box[0] + box[2] / 2) * dw, (box[1] + box[3] / 2) * dh, box[2] * dw, box[3] * dh
  10.  
  11.       (dir / 'labels').mkdir(parents=True, exist_ok=True)  # make labels directory
  12.       pbar = tqdm((dir / 'annotations').glob('*.txt'), desc=f'Converting {dir}')
  13.       for f in pbar:
  14.           img_size = Image.open((dir / 'images' / f.name).with_suffix('.jpg')).size
  15.           lines = []
  16.           with open(f, 'r') as file:  # read annotation.txt
  17.               for row in [x.split(',') for x in file.read().strip().splitlines()]:
  18.                   if row[4] == '0':  # VisDrone 'ignored regions' class 0
  19.                       continue
  20.                   cls = int(row[5]) - 1
  21.                   box = convert_box(img_size, tuple(map(int, row[:4])))
  22.                   lines.append(f"{cls} {' '.join(f'{x:.6f}' for x in box)}\n")
  23.                   with open(str(f).replace(os.sep + 'annotations' + os.sep, os.sep + 'labels' + os.sep), 'w') as fl:
  24.                       fl.writelines(lines)  # write label.txt

但是因为在白天使用AutoDL这个平台的时候，下载的过程里，AutoDL连接外网十分不稳定，一会可以连接github外网，一会又连不上github，下载时断时续，导致VisDrone.yaml文件里面的下载步骤被中止。

下载一旦被迫中止，后续的原始数据进行加工的步骤自然也没来得及运行。

当时当你，此时再次运行train.py的时候，代码会识别你是否已经有几个下载下来的文件。只要有一个文件被下载下来，就不会继续之前的下载，最关键的是代码会同时自动跳过数据加工这个步骤。

此后无论你执行“python train.py”这句代码多少次，都不会 把缺少的数据压缩包下载下来，同时也不会将 annotations文件夹进行再加工，也就没有labels这个文件夹，自然就会不停的报下面这个错，说找不到labels.cache这个文件。无论你执行“python train.py”这句代码多少次，都还是下面这个报错。我曾经为了这个报错，苦恼了一下午，怎么样debug都修不好。截图留念

  1. train: Scanning /root/m_coding/yolov3/datasets/VisDrone/datasets/VisDrone/VisDrone2019-DET-train/labels.cache... 0 images, 6471 backgrounds, 0 
  2. Traceback (most recent call last):
  3.   File "train.py", line 3, in <module>
  4.     model.train(data="VisDrone.yaml", epochs=1,batch=1)  
  5.   File "/root/m_coding/YOLOv8n+LADHead/ultralytics/yolo/engine/model.py", line 251, in train
  6.     self.trainer.train()
  7.   File "/root/m_coding/YOLOv8n+LADHead/ultralytics/yolo/engine/trainer.py", line 187, in train
  8.     self._do_train(RANK, world_size)
  9.   File "/root/m_coding/YOLOv8n+LADHead/ultralytics/yolo/engine/trainer.py", line 252, in _do_train
  10.     self._setup_train(rank, world_size)
  11.   File "/root/m_coding/YOLOv8n+LADHead/ultralytics/yolo/engine/trainer.py", line 237, in _setup_train
  12.     self.train_loader = self.get_dataloader(self.trainset, batch_size=batch_size, rank=rank, mode="train")
  13.   File "/root/m_coding/YOLOv8n+LADHead/ultralytics/yolo/v8/detect/train.py", line 42, in get_dataloader
  14.     build_dataloader(self.args, batch_size, img_path=dataset_path, stride=gs, rank=rank, mode=mode,
  15.   File "/root/m_coding/YOLOv8n+LADHead/ultralytics/yolo/data/build.py", line 71, in build_dataloader
  16.     dataset = YOLODataset(
  17.   File "/root/m_coding/YOLOv8n+LADHead/ultralytics/yolo/data/dataset.py", line 44, in __init__
  18.     super().__init__(img_path, imgsz, cache, augment, hyp, prefix, rect, batch_size, stride, pad, single_cls)
  19.   File "/root/m_coding/YOLOv8n+LADHead/ultralytics/yolo/data/base.py", line 49, in __init__
  20.     self.labels = self.get_labels()
  21.   File "/root/m_coding/YOLOv8n+LADHead/ultralytics/yolo/data/dataset.py", line 114, in get_labels
  22.     raise FileNotFoundError(f"{self.prefix}No labels found in {cache_path}, can not start training. {HELP_URL}")
  23. FileNotFoundError: train: No labels found in /root/m_coding/yolov3/datasets/VisDrone/datasets/VisDrone/VisDrone2019-DET-train/labels.cache, can not start training. See https://github.com/ultralytics/yolov5/wiki/Train-Custom-Data

总结来说，使用YOLOv8的时候，下载数据集就要一气呵成，中间不能出现任何网络中断。如果下载中断了，删除掉所有你已经下载好的所有文件，从头开始下载。多次下载，直到你碰到下载整个过程中网络通畅，一口气下载下来，然后让代码把数据加工完。

        至于外网github的连接时断时续，有可能是因为，白天的时候有许多人通过AutoDL的这个外网VPN访问github，但是AutoDL提供的VPN的带宽有限，导致你分配到的下载带宽特别窄。所以你可以尝试在半夜的时候运行训练程序，把数据下载和加工好。

讲解如何解决这个报错

方向1：删除掉你已经下载的数据集，从头下载，直到可以流畅下载并加工annotations数据

这个方法是最有效、最高效、最切实可用的方法。

重复这个方法多次，直到你也能像我下面这样，可以把下面四个文件流畅的下载下来，并解压好

VisDrone2019-DET-test-challenge, VisDrone2019-DET-test-dev，VisDrone2019-DET-train，VisDrone2019-DET-val

你的terminal里面一定要有下面这三行，这三行表示，代码成功将“annotations”文件夹里面的数据加工成了YOLOv8指定的数据格式，形成了labels文件夹

  1. Converting /root/m_coding/yolov3/datasets/VisDrone/datasets/VisDrone/VisDrone2019-DET-train: 6471it [04:42, 22.88it/s] 
  2. Converting /root/m_coding/yolov3/datasets/VisDrone/datasets/VisDrone/VisDrone2019-DET-val: 548it [00:31, 17.59it/s]
  3. Converting /root/m_coding/yolov3/datasets/VisDrone/datasets/VisDrone/VisDrone2019-DET-test-dev: 1610it [00:59, 27.17it/s]
  4. Dataset download success ✅ (574.6s), saved to /root/m_coding/yolov3/datasets/VisDrone/datasets

方向2：拿现成的labels文件夹过来用

如果你之前保存了 traIn dev 和test三个数据的  制作加工好的label文件夹和label.cache缓存文件。你可以直接放到这三个文件夹里面，放好以后是下面这样

train

val 验证集

test-dev 公开label的测试集

方向3：搞定下载的网络 

问客服，为什么github的网络连接多次无法访问？要求他修复网络，保证下载数据过程中，与github之间的连接是稳定的

如果github的网络连接修复好了，保证下载的时候不间断，完整的把所有数据在一次运行的时候下载下来，

再运行一次，看看能不能正常制作label

方向4：将本地数据集上传，然后自己制作label文件夹

如果你电脑上，没有VisDrone 训练 测试 验证 三个数据集的labels这个加工后的这个标注数据文件夹，你可以私信问我要。你也可以，绕开执行VisDrone.yaml这条途径，转而通过下面这个方法，自己制作出来。

通过这个路径“/ultralytics/yolo/data/datasets/VisDrone.yaml”可以找到YOLOv8的 VisDrone.yaml 数据集配置文件。

内容如下

  1. # Ultralytics YOLO 
    声明：本文内容由网友自发贡献，不代表【wpsshop博客】立场，版权归原作者所有，本站不承担相应法律责任。如您发现有侵权的内容，请联系我们。转载请注明出处：https://www.wpsshop.cn/w/Cpp五条/article/detail/618039
    推荐阅读
    相关标签

Copyright © 2003-2013 www.wpsshop.cn 版权所有，并保留所有权利。