Dataset · eltorio
79.8k radiology images with captions
eltorio/ROCOv2-radiology
Multimodal set of radiology scans with captions and medical concepts from PMC.
What's inside
ROCOv2 is a multimodal dataset of radiology images, captions, and medical concepts from the PMC Open Access Subset. An updated version of ROCO, it adds 35,705 images and improves concept extraction and filtering. Totaling 79,789 images, each with a caption and medical concepts; English language.
What you can do with it
- Training image-text-to-text in radiology
- Fine-tuning VLMs for scan captioning
- Cross-modal image-text retrieval in medicine
- Languages
- English
- Size
- 10K–100K rows
- Downloads this month
- 13.6K
- Likes
- 91
Models trained on it
- WafaaFraih/blip-roco-radiology-captioningImage → text 66
- WafaaFraih/medical-image-captioning-roco 29
- Ozantsk/biomedclip-rocov2-finetuned 16
- erfanasghariyan/resnet50_roco_radiocaptioner_big 10
- eltorio/IDEFICS3_ROCOv2Image text → text 8
- bouthros/llama32_11b_vision_medical_finetune 0
- BoghdadyJR/qwen_reportImage text → text 0
- siva317/llama32-vision-radiology-lora 0
How to load it
from datasets import load_dataset
ds = load_dataset("eltorio/ROCOv2-radiology")