datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
legacysurvey_hsc_crossmatched
Crossmatched samples from the Multimodal Universe for Legacy Survey/HSC
Mother paper here: https://arxiv.org/abs/2412.02527
legacysurvey_hsc_crossmatched
Crossmatched samples from the Multimodal Universe for Legacy Survey/HSC
Mother paper here: https://arxiv.org/abs/2412.02527
hsc-anomaly-expert
HSC Anomaly Expert
This dataset contains 1,000 galaxy images independently scored by 16 expert
astronomers. The ten fixed few-shot examples form the train split; the
remaining 990 images form the test split used for evaluation, containing
45 expert-consensus anomalies and 945 other images.
The images come from public data release 2 of the Subaru Hyper Suprime-Cam
(HSC) survey; this is the HSC in the dataset name. The source images,
Zooniverse subject metadata, volunteer… See the full description on the dataset page: https://huggingface.co/datasets/astronolan/hsc-anomaly-expert.bd-hsc-multimodalcosmosweb-hsc-jwst-high-snr-pilMonitorDigitscosmosweb-hsc-jwst-high-snr-pil2hsc-bd-mmcosmosweb-hsc-jwst-high-snr-pil-sigmoidcosmosweb-hsc-jwst-high-snr-pil2eurosat-imagesSA-VLM-HarmImages
Malicious Image Dataset
This dataset is the dataset for Malicious Image Analysis via Vision-Language Segmentation Fusion: Detection, Element, and Location in One-shot.
Dataset Overview
The dataset contains various categories of harmful images, such as:
Violence
Illegal Activities
Extremism and Discrimination
Each image is provided with corresponding annotation files for segmentation and classification tasks.
Legal Notice on Porn Category
Due to legal… See the full description on the dataset page: https://huggingface.co/datasets/HS-Coriander/SA-VLM-HarmImages.