datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
food-nutrients
Food Nutrients: A Macronutrients Dataset
food_nutrients is a dataset of visual and nutritional data for ~3k realistic plates of food captured from Google cafeterias using a custom scanning rig.
This dataset provides annotated pictures of food plates along with calories, macronutrients (fat, carbohydrate, protein) for the total plate and for every ingredient as well.
Column
Definition
image
a 640x640 top-down image from a realistic food plate in the Google cafeteria… See the full description on the dataset page: https://huggingface.co/datasets/Sebastianrtj/food-nutrients.form-field-v1-benchmark
form-field-v1-benchmark
The evaluation set behind the form-field-v1 form-field detectors and the
leaderboard — 5,914 form pages with
126,938 annotated fields (Text, Choice = checkbox/radio, Signature), across three variants:
variant
images
fields
notes
empty
2,334
49,171
blank forms
filled
1,826
39,797
digitally-typed values
handwritten
1,754
37,970
handwritten (ink) values
A Nutrient dataset, released under CC-BY-4.0.
🎯 Demo ·
🏆 Leaderboard… See the full description on the dataset page: https://huggingface.co/datasets/nutrientdocs/form-field-v1-benchmark.food-nutrients
Food Nutrients: A Macronutrients Dataset
food_nutrients is a dataset of visual and nutritional data for ~3k realistic plates of food captured from Google cafeterias using a custom scanning rig.
This dataset provides annotated pictures of food plates along with calories, macronutrients (fat, carbohydrate, protein) for the total plate and for every ingredient as well.
Column
Definition
image
a 640x640 top-down image from a realistic food plate in the Google cafeteria
id… See the full description on the dataset page: https://huggingface.co/datasets/mmathys/food-nutrients.chart-parsing-benchmark
Chart Parsing Benchmark
Chart image in, structured JSON out. A fixed, test-only benchmark for turning a chart image into its
title, chart type, data table, encoding, axes, series, legend, and data labels. It holds
2500 charts and is the set behind the leaderboard. Every entry sees the same charts and the same
prompt, and the reference scorer ships in this repository, so any result here can be reproduced.
🎯 Model: nutrientdocs/chart-parsing-vlm
🧪 Try it:… See the full description on the dataset page: https://huggingface.co/datasets/nutrientdocs/chart-parsing-benchmark.nutrient-detection-layout
Nutrient extraction dataset
This dataset contains annotated images of nutrition tables. The goal of this dataset was to train a model to extract nutrient values from nutrition tables, as part of the Nutrisight project.
It contains ~3k samples in total (2.8k for training and 199 for testing). For more information about the project, please refer to the nutrisight directory in the openfoodfacts-ai GitHub repository.
The images were collected from the Open Food Facts database, and… See the full description on the dataset page: https://huggingface.co/datasets/openfoodfacts/nutrient-detection-layout.form-field-vlm-v2-benchmark
form-field-vlm-v2-benchmark
A balanced held-out slice for evaluating form-field detect + extract — detection + fine type + text
label + entered value — across the three render conditions a real pipeline meets: empty, filled
(printed), and handwritten. Drawn from the commonforms-synth-v2 split (document-disjoint from training).
📊 Benchmark: nutrientdocs/form-field-vlm-v2-benchmark
🎯 Model: nutrientdocs/form-field-vlm-v2 · 🏆 Leaderboard
Composition
354 pages —… See the full description on the dataset page: https://huggingface.co/datasets/nutrientdocs/form-field-vlm-v2-benchmark.doc-split-benchmark
Doc-Split Benchmark
The evaluation slice for page-stream segmentation — the exact set behind the
leaderboard and the cloud-VLM
comparison. Self-contained (page images embedded), with a reference scorer so results are reproducible.
This is the benchmark, not the training corpus (which stays private).
🏆 Leaderboard: doc-split-leaderboard
🎯 Demo: doc-split-demo
🟢 Model: doc-split-mini-e5 (open weights)
🌍 OpenPSS cuts: openpss-mirror (SHORT/LONG, self-contained)… See the full description on the dataset page: https://huggingface.co/datasets/nutrientdocs/doc-split-benchmark.document-classification-benchmark
Document Classification Benchmark (open-vocab, zero-shot)
Given a document image and an arbitrary set of text labels, which one is right? A held-out, zero-shot,
open-vocabulary evaluation for document-type classification — labels are supplied at inference, not baked
into a head. Test split only; not for training. Every image is drawn from a permissively-licensed,
redistributable source.
Powers the
document-classification-leaderboard
and evaluates document-classification-v2… See the full description on the dataset page: https://huggingface.co/datasets/nutrientdocs/document-classification-benchmark.food-nutrients
Food Nutrients: A Macronutrients Dataset
food_nutrients is a dataset of visual and nutritional data for ~3k realistic plates of food captured from Google cafeterias using a custom scanning rig.
This dataset provides annotated pictures of food plates along with calories, macronutrients (fat, carbohydrate, protein) for the total plate and for every ingredient as well.
Column
Definition
image
a 640x640 top-down image from a realistic food plate in the Google cafeteria
id… See the full description on the dataset page: https://huggingface.co/datasets/sunli1201/food-nutrients.form-field-vlm-benchmark
Form Field Detection Benchmark
A fixed, test-only benchmark for end-to-end form-field extraction. It contains the exact 100 clean
form pages and 701 fields used by the
form-field-vlm scorecard and
leaderboard. Try the complete
server-side pipeline in the demo.
The task is to return every interactive widget as constrained JSON with:
box: [x0, y0, x1, y1] on a 0–1000 page grid
type: text, choice_checkbox, choice_radio, choice_select, or signature
label: the field's text label… See the full description on the dataset page: https://huggingface.co/datasets/nutrientdocs/form-field-vlm-benchmark.doc-openvocab-benchmark
Open-Vocab Document & Figure Classification Benchmark
Given a document or figure image and an arbitrary set of text labels, which one is right? This is a
zero-shot, open-vocabulary image-classification benchmark for the document-AI setting: every image is
scored against a broad ~48-label candidate vocabulary (document types + figure/zone types), and the task
is to pick the correct label. The labels are supplied at inference — which is precisely what a fixed-label
supervised… See the full description on the dataset page: https://huggingface.co/datasets/nutrientdocs/doc-openvocab-benchmark.openpss-mirror
OpenPSS — community mirror
⚠️ This is a redistribution (mirror) of the OpenPSS benchmark, not our own work. It is hosted for
availability and reproducibility. All credit belongs to the original authors. If you are an author or
rights-holder and would like any change or removal, please open a discussion here or contact us.
Original work
OpenPSS: An Open Page Stream Segmentation Benchmark — Ruben van Heusden, Jaap Kamps, Maarten Marx
(University of Amsterdam… See the full description on the dataset page: https://huggingface.co/datasets/nutrientdocs/openpss-mirror.banana_leaf_nutrient_classification
Banana Leaf Nutrient Classification
A dataset for classification of Banana leaf nutrient deficiencies. The dataset contains raw and augmented versions.The raw dataset contains 5,348 images.Images per class:
Boron: 173
Calcium: 794
Healthy: 1,584
Iron: 151
Magnesium: 288
Manganese: 24
Potassium: 381
Sulphur: 1,240
Zinc: 713
The augmented dataset contains 12,747 images.Images per class:
Boron: 1,384
Calcium: 1,588
Healthy: 1,586
Iron: 1,359
Magnesium: 1,440
Manganese: 1,200… See the full description on the dataset page: https://huggingface.co/datasets/Project-AgML/banana_leaf_nutrient_classification.nutrient5k-test-100-depthnutrient5k-test-100food-nutrients-preparated