datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
Recruitment-Task-3
DeepWeeds - AI-MED AGH convenience mirror
This is a convenience mirror of the official DeepWeeds image archive and the
upstream annotations pinned to a specific commit. original/images.zip is
preserved unchanged; images are not extracted or duplicated here. models.zip
from the source authors is deliberately not mirrored.
Dataset facts
17,509 in-situ images from Queensland, Australia.
Nine classes: eight weed species plus Negative.
The authors publish five folds… See the full description on the dataset page: https://huggingface.co/datasets/AI-MED-AGH/Recruitment-Task-3.aimeghamuseRecruitment-Task-2A
BloodMNIST - AI-MED AGH convenience mirror
This repository is an AI-MED AGH convenience mirror of the unchanged official
BloodMNIST NPZ distribution from the MedMNIST project.
The original bloodmnist.npz is preserved byte-for-byte in original/.
Dataset facts
28x28 RGB images with eight blood-cell classes.
Official splits: 11,959 training images, 1,712 validation images, and 3,421 test images.
Classes:
basophil
eosinophil
erythroblast
immature granulocytes… See the full description on the dataset page: https://huggingface.co/datasets/AI-MED-AGH/Recruitment-Task-2A.AIME-Dataset
AIME: AI Multimedia Ethics Dataset
The AIME (AI Multimedia Ethics) dataset is a collection of images and videos generated by
Text-to-Image (T2I) and Text-to-Video (T2V) models, manually annotated for ethical/unethical content.
Access
This dataset is gated. To access it, please request approval on this page.Access is granted for non-commercial research purposes only.
Dataset Summary
The AIME dataset was created to study the risks of generative AI… See the full description on the dataset page: https://huggingface.co/datasets/DAISLab-Unisa/AIME-Dataset.