Datasets:

marcin119a
/

polish-traffic-12k

Exception:    SplitsNotFoundError
Message:      The split names could not be parsed from the dataset config.
Traceback:    Traceback (most recent call last):
                File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 286, in get_dataset_config_info
                  for split_generator in builder._split_generators(
                                         ~~~~~~~~~~~~~~~~~~~~~~~~~^
                      StreamingDownloadManager(base_path=builder.base_path, download_config=download_config)
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                  )
                  ^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/webdataset/webdataset.py", line 80, in _split_generators
                  raise ValueError(
                  ...<2 lines>...
                  )
              ValueError: The TAR archives of the dataset should be in WebDataset format, but the files in the archive don't share the same prefix or the same types.
              
              The above exception was the direct cause of the following exception:
              
              Traceback (most recent call last):
                File "/src/services/worker/src/worker/job_runners/config/split_names.py", line 68, in compute_split_names_from_streaming_response
                  for split in get_dataset_split_names(
                               ~~~~~~~~~~~~~~~~~~~~~~~^
                      path=dataset,
                      ^^^^^^^^^^^^^
                      config_name=config,
                      ^^^^^^^^^^^^^^^^^^^
                      token=hf_token,
                      ^^^^^^^^^^^^^^^
                  )
                  ^
                File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 340, in get_dataset_split_names
                  info = get_dataset_config_info(
                      path,
                  ...<6 lines>...
                      **config_kwargs,
                  )
                File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 291, in get_dataset_config_info
                  raise SplitsNotFoundError("The split names could not be parsed from the dataset config.") from err
              datasets.inspect.SplitsNotFoundError: The split names could not be parsed from the dataset config.

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

Traffic Road Object Detection Polish 12k

Dashcam frames from Polish roads with bounding boxes in YOLO format (11 classes: cars, pedestrians, traffic lights, several kinds of traffic signs).

This is a re-upload of the Kaggle dataset mikoajkoek/traffic-road-object-detection-polish-12k (author: mikoajkoek, license: Apache 2.0), repackaged for fast download.

Changes with respect to the original

Images and labels are grouped into .tar shards (about 1 GB each), because the original folders hold more than 10 000 files each, which is slow to download from the Hub.
Added data.yaml with class names. The original archive has none, so the order was recovered from the per-class counts in stats.csv (alphabetical order, as exported by Roboflow) and checked visually.
Added stats.csv (the original data.csv). No images or labels were modified.

Layout

Split	Files (images)	Shards
`train`	10 474	`data/train-*.tar`
`valid`	935	`data/valid-*.tar`
`test`	484	`data/test-*.tar`

Inside the shards the paths are <split>/images/<name> and <split>/labels/<name>.txt, so extracting all of them into one directory gives the standard YOLO layout described by data.yaml.

hf download <repo_id> --repo-type dataset --local-dir polish12k
cd polish12k && for f in data/*.tar; do tar xf "$f"; done   # -> train/ valid/ test/

Classes

0 Car, 1 Different-Traffic-Sign, 2 Green-Traffic-Light, 3 Motorcycle, 4 Pedestrian, 5 Pedestrian-Crossing, 6 Prohibition-Sign, 7 Red-Traffic-Light, 8 Speed-Limit-Sign, 9 Truck, 10 Warning-Sign.

Caveats

Augmented copies. Each split mixes original .jpg frames with Roboflow-augmented .png copies (file names contain _aug_out_), also in valid and test. Near-duplicates inflate evaluation scores, so for validation and testing use only the files without _aug_out_.
Frames come from a small number of videos, so neighbouring frames are very similar.
Class Motorcycle is rare (a few dozen train instances in the original, non-augmented set).
The boxes cover only these 11 classes. Other objects (e.g. cyclists, buses) are not labelled and are background in this dataset.
Privacy: the source frames appear to have blurred license plates (spot check of cars from 15 different videos, plus a few full frames). Not every image was reviewed, and faces were not checked.

License and attribution

Apache License 2.0 (see LICENSE). Original dataset by mikoajkoek on Kaggle, linked above.

Downloads last month: 33

Total file size:

10.9 GB