Dataset · visual-layer
Enriched ImageNet-1K with descriptions and object boxes
visual-layer/imagenet-1k-vl-enriched
An augmented version of ImageNet-1K with over a million records containing text descriptions, bounding boxes, and label error notes.
What's inside
This is an enriched version of the classic ImageNet-1K dataset, supplemented with detailed English text descriptions, object bounding boxes, and information on labeling issues. The set contains from 1 to 10 million records organized in a 6-column table, including original file names. Error notes allow filtering out noise and assembling a cleaner sample.
What you can do with it
- Training multimodal models for image-to-text, text-to-image, and visual question answering tasks
- Training object detectors and cleaning classic datasets from labeling errors
- Tasks
- Object detection, Image classification, Text → image, Image → text, Visual question answering
- Languages
- English
- Size
- 1M–10M rows
- Downloads this month
- 3.1K
- Likes
- 41
Models trained on it
- Leonardo6/sft-llava-v1.6-vicuna-7b-hfImage text → text 24
- Leonardo6/llava-sft-freezeImage text → text 15
- Leonardo6/sft-llava-1.5-7b-hfImage text → text 13
- mascIT/gradientlab-exp20251025_0_vlm_20m_in1k 7
- Leonardo6/llava-sftImage text → text 6
- fakekungfu/Fusion-Prior-H14_B32_VAE 6
- hkchengrex/c2ot-imagenet32-fmText → image 0
- hkchengrex/c2ot-imagenet32-otText → image 0
How to load it
from datasets import load_dataset
ds = load_dataset("visual-layer/imagenet-1k-vl-enriched")