datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
Signature-Verification-Dataset
Multilingual Signature Verification Dataset
Dataset Summary
The Multilingual Signature Verification Dataset is a curated collection of handwritten signatures designed for offline signature verification and related computer vision tasks.
The dataset contains more than 7,000 signature images spanning three major writing systems:
Hindi
Bengali
English
The English portion includes samples from the well-known CEDAR Signature Dataset, while additional Hindi and… See the full description on the dataset page: https://huggingface.co/datasets/rakshitdabral/Signature-Verification-Dataset.flippd-depop-verify
Flippd Depop benchmark — public verification subset
Anonymized subset for independently reproducing the Flippd Depop leave-X-out
recommendation benchmark without model weights or the full dataset.
metadata.jsonl — one row per listing: img_id, seller_id (anonymized),
gender, category, color0, brand, price.
embeddings.npz — precomputed vectors (model outputs, not weights) for
every encoder scored in the study: resnet, clip, fc_nohead, fc_head
(the study's private encoder)… See the full description on the dataset page: https://huggingface.co/datasets/rayna-rules/flippd-depop-verify.Signature-Verification-Dataset
Multilingual Signature Verification Dataset
Dataset Summary
The Multilingual Signature Verification Dataset is a curated collection of handwritten signatures designed for offline signature verification and related computer vision tasks.
The dataset contains more than 7,000 signature images spanning three major writing systems:
Hindi
Bengali
English
The English portion includes samples from the well-known CEDAR Signature Dataset, while additional Hindi and… See the full description on the dataset page: https://huggingface.co/datasets/Peachyy2208/Signature-Verification-Dataset.flippd-verify
Flippd benchmark — public verification subset
Anonymized subset for independently reproducing the Flippd leave-X-out
recommendation benchmark without model weights or the full dataset.
metadata.jsonl — one row per listing: img_id, seller_id (anonymized),
gender, category, color0, brand, price.
embeddings.npz — precomputed vectors (model outputs, not weights) for
every encoder scored in the study: resnet, clip, fc_nohead, fc_head
(the study's private encoder), aligned to ids.… See the full description on the dataset page: https://huggingface.co/datasets/rayna-rules/flippd-verify.saffron-verify
SaffronVerify Dataset
Dataset Summary
SaffronVerify is a fine-grained image classification dataset for saffron quality grading. It contains images of saffron across three quality categories — from premium grade to adulterated samples — intended for training computer vision models to detect saffron purity and adulteration.
Dataset Structure
The dataset follows the standard ImageFolder layout and is split into training and validation sets.
saffron-verify/
├──… See the full description on the dataset page: https://huggingface.co/datasets/Arko007/saffron-verify.