Hugging Face's logo

Datasets:
juliozhao
/
DocSynth300K

Modalities:
Text
Formats:
parquet
Size:
100K - 1M
ArXiv:
Libraries:
Datasets
Dask
Dataset card Viewer Files Community
1
Dataset Viewer (First 5GB)
Auto-converted to Parquet
Full Screen
filename
stringlengths
18
22
image_data
unknown
anno_string
sequencelengths
1
32
1720629091_634364.jpg
"/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAgGBgcGBQgHBwcJCQgKDBQNDAsLDBkSEw8UHRofHh0aHBwgJC4nICIsIxwcKDcpLDA(...TRUNCATED)
["23 0.09418776004975932 0.5595031058841483 0.7866484741599264 0.5595031058841483 0.7866484741599264(...TRUNCATED)
1720631218_0076387.jpg
"/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAgGBgcGBQgHBwcJCQgKDBQNDAsLDBkSEw8UHRofHh0aHBwgJC4nICIsIxwcKDcpLDA(...TRUNCATED)
["48 0.3203970388133509 0.38638961743071687 0.8027139596682623 0.38638961743071687 0.802713959668262(...TRUNCATED)
1720630536_4433818.jpg
"/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAgGBgcGBQgHBwcJCQgKDBQNDAsLDBkSEw8UHRofHh0aHBwgJC4nICIsIxwcKDcpLDA(...TRUNCATED)
["23 0.20592878012748345 0.5547806696868959 0.3459488196369184 0.5547806696868959 0.3459488196369184(...TRUNCATED)
1720628932_9465413.jpg
"/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAgGBgcGBQgHBwcJCQgKDBQNDAsLDBkSEw8UHRofHh0aHBwgJC4nICIsIxwcKDcpLDA(...TRUNCATED)
["48 0.7416247650173948 0.13510898025034856 0.9638921098434892 0.13510898025034856 0.963892109843489(...TRUNCATED)
1720626855_443507.jpg
"/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAgGBgcGBQgHBwcJCQgKDBQNDAsLDBkSEw8UHRofHh0aHBwgJC4nICIsIxwcKDcpLDA(...TRUNCATED)
["48 0.22310174015164067 0.8383290297078849 0.40056791527089386 0.8383290297078849 0.400567915270893(...TRUNCATED)
1720629671_196503.jpg
"/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAgGBgcGBQgHBwcJCQgKDBQNDAsLDBkSEw8UHRofHh0aHBwgJC4nICIsIxwcKDcpLDA(...TRUNCATED)
["48 0.20161389936111906 0.7887753765058141 0.5474765563362877 0.7887753765058141 0.5474765563362877(...TRUNCATED)
1720627929_9200647.jpg
"/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAgGBgcGBQgHBwcJCQgKDBQNDAsLDBkSEw8UHRofHh0aHBwgJC4nICIsIxwcKDcpLDA(...TRUNCATED)
["48 0.056336793813028696 0.41674915120716266 0.9389167182217064 0.41674915120716266 0.9389167182217(...TRUNCATED)
1720625738_4356627.jpg
"/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAgGBgcGBQgHBwcJCQgKDBQNDAsLDBkSEw8UHRofHh0aHBwgJC4nICIsIxwcKDcpLDA(...TRUNCATED)
["23 0.10393390289434445 0.3474227825138238 0.6252506620550092 0.3474227825138238 0.6252506620550092(...TRUNCATED)
1720629638_6551414.jpg
"/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAgGBgcGBQgHBwcJCQgKDBQNDAsLDBkSEw8UHRofHh0aHBwgJC4nICIsIxwcKDcpLDA(...TRUNCATED)
["48 0.13555293884769376 0.7020600531831678 0.2602049533137642 0.7020600531831678 0.2602049533137642(...TRUNCATED)
1720626221_2315247.jpg
"/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAgGBgcGBQgHBwcJCQgKDBQNDAsLDBkSEw8UHRofHh0aHBwgJC4nICIsIxwcKDcpLDA(...TRUNCATED)
["48 0.5480442607393465 0.7865330724474383 0.6467595692656398 0.7865330724474383 0.6467595692656398 (...TRUNCATED)
End of preview. Expand in Dataset Viewer.
YAML Metadata Warning: empty or missing yaml metadata in repo card (https://huggingface.co/docs/hub/datasets-cards)

image/png

DocSynth300K is a large-scale and diverse document layout analysis pre-training dataset, which can largely boost model performance.

Data Download

Use following command to download dataset(about 113G):

from huggingface_hub import snapshot_download
# Download DocSynth300K
snapshot_download(repo_id="juliozhao/DocSynth300K", local_dir="./docsynth300k-hf", repo_type="dataset")
# If the download was disrupted and the file is not complete, you can resume the download
snapshot_download(repo_id="juliozhao/DocSynth300K", local_dir="./docsynth300k-hf", repo_type="dataset", resume_download=True)

Data Formatting & Pre-training

If you want to perform DocSynth300K pretraining, using format_docsynth300k.py to convert original .parquet format into YOLO format. The converted data will be stored at ./layout_data/docsynth300k.

python format_docsynth300k.py

To perform DocSynth300K pre-training, use this command. We default use 8GPUs to perform pretraining. To reach optimal performance, you can adjust hyper-parameters such as imgsz, lr according to your downstream fine-tuning data distribution or setting.

Note: Due to memory leakage in YOLO original data loading code, the pretraining on large-scale dataset may be interrupted unexpectedly, use --pretrain last_checkpoint.pt --resume to resume the pretraining process.

https://huggingface.co/papers/2410.12628

Downloads last month
347
Edit dataset card
Size of the auto-converted Parquet files (First 5GB):
6.12 GB
Number of rows (First 5GB):
15,311
Estimated number of rows:
229,033

Models trained or fine-tuned on juliozhao/DocSynth300K