What causes "VOCdevkit dose not in path:'{}'." in WZMIAOMIAO/deep-learning-for-image-processing?

Pass the parent directory that directly contains `VOCdevkit` via --data-path

WZMIAOMIAO/deep-learning-for-image-processing · error · FileNotFoundError

VOCdevkit dose not in path:'{}'.

Error message

VOCdevkit dose not in path:'{}'.

What it means

The multi-GPU FCN training script checks that `<args.data_path>/VOCdevkit` exists before building datasets. If the directory is missing it raises FileNotFoundError, because VOCSegmentation would otherwise fail later with a more confusing path error. It is an early environment/data-layout guard.

Source

Thrown at pytorch_segmentation/fcn/train_multi_GPU.py:84

    return model


def main(args):
    init_distributed_mode(args)
    print(args)

    device = torch.device(args.device)
    # segmentation nun_classes + background
    num_classes = args.num_classes + 1

    # 用来保存coco_info的文件
    results_file = "results{}.txt".format(datetime.datetime.now().strftime("%Y%m%d-%H%M%S"))

    VOC_root = args.data_path
    # check voc root
    if os.path.exists(os.path.join(VOC_root, "VOCdevkit")) is False:
        raise FileNotFoundError("VOCdevkit dose not in path:'{}'.".format(VOC_root))

    # load train data set
    # VOCdevkit -> VOC2012 -> ImageSets -> Segmentation -> train.txt
    train_dataset = VOCSegmentation(args.data_path,
                                    year="2012",
                                    transforms=get_transform(train=True),
                                    txt_name="train.txt")
    # load validation data set
    # VOCdevkit -> VOC2012 -> ImageSets -> Segmentation -> val.txt
    val_dataset = VOCSegmentation(args.data_path,
                                  year="2012",
                                  transforms=get_transform(train=False),
                                  txt_name="val.txt")

    print("Creating data loaders")
    if args.distributed:
        train_sampler = torch.utils.data.distributed.DistributedSampler(train_dataset)
        test_sampler = torch.utils.data.distributed.DistributedSampler(val_dataset)

View on GitHub (pinned to 1ec3fe6f37)

Solutions

Pass the parent directory that directly contains `VOCdevkit` via --data-path
Verify layout: data_path/VOCdevkit/VOC2012/ImageSets/Segmentation/train.txt exists
Use an absolute path for --data-path when launching torchrun from another CWD
Check mount points/volume permissions if training inside a container

Example fix

// before
python train_multi_GPU.py --data-path /home/data/VOCdevkit   # wrong: nested too deep
// after
python train_multi_GPU.py --data-path /home/data             # contains VOCdevkit/

Defensive patterns

Strategy: validation

Validate before calling

import os
data_path = args.data_path
assert os.path.isdir(os.path.join(data_path, 'VOCdevkit', 'VOC2012', 'ImageSets', 'Segmentation')), \
    f"VOC dataset layout missing under {data_path}"

Type guard

def voc_root_ok(path):
    return os.path.isdir(os.path.join(path, 'VOCdevkit'))

Try / catch

try:
    train_dataset = VOCSegmentation(args.data_path, year="2012", ...)
except FileNotFoundError as e:
    sys.exit(f"Dataset not found at {args.data_path}: {e}. Download/extract VOC2012 first.")

Prevention

Always pass the directory that contains VOCdevkit, not VOCdevkit itself
Use absolute paths for --data-path
Verify dataset layout on every node in multi-GPU/multi-node jobs
Add a dataset download/extract step to your launch script

When it happens

Trigger: Running `train_multi_GPU.py` with `--data-path` pointing to a directory that does not contain a `VOCdevkit` folder; passing the VOCdevkit folder itself instead of its parent; extracting the dataset elsewhere; typo in the path or relative path resolved from the wrong CWD.

Common situations: Distributed training launched from a different working directory so relative paths break; downloading only VOC2012 images but not the SegmentationClass annotations layout; pointing at COCO root by mistake; Docker mounts missing the dataset.

Related errors

AI-assisted analysis of WZMIAOMIAO/deep-learning-for-image-processing@1ec3fe6f37 (2026-08-30). Data as JSON: /api/errors/274afff6760a566c. Report an issue: GitHub.