Open Datasets

Browse 500+ open source datasets for your next machine learning project. Our list of free datasets keeps growing, so make sure you visit it frequently.

Oops! Something went wrong while submitting the form.
Filters
Use Cases
Clear
TVQA Dataset
TVQA Dataset
A Localized, Compositional Video Question Answering Dataset
Visual Question Answering
Bounding Boxes
310800
Items
Classes
Labels
YouTube-VOS
YouTube-VOS
A Large-Scale Benchmark for Video Object Segmentation
Video Object Segmentation
Semantic Segmentation
YouTube-VOS
190000
Items
90
Classes
190000
Labels
HDM05
HDM05
HDM05
3D Object Pose
Keypoint Skeleton
Universität Bonn
Items
Classes
Labels
SPARE3D
SPARE3D
A Dataset for SPAtial REasoning on Three-View Line Drawings
3D Object Detection
Bounding Boxes
New York University
Items
Classes
Labels
Chalearn CASIA-SURF Dataset
Chalearn CASIA-SURF Dataset
Large-scale face Anti-spoofing dataset
Face Detection
Bounding Boxes
21000
Items
1000
Classes
21000
Labels
Animal-Pose Dataset
Animal-Pose Dataset
Dataset with animal pose annotations
Object Detection
Bounding Boxes
Shanghai Jiao Tong University
20
Items
5
Classes
4000
Labels
SKU-110K
SKU-110K
110k categories of densely packed items in store
Object Detection
Bounding Boxes
110000
Items
110000
Classes
110000
Labels
CelebA-Spoof
CelebA-Spoof
Large-scale face anti-spoofing dataset
Face Recognition
Bounding Boxes
Beijing Jiaotong University, Beijing, China
625537
Items
10177
Classes
625537
Labels
CheXpert
CheXpert
A Large Chest X-Ray Dataset And Competition
Medical Images
Semantic Segmentation
Stanford ML Group
200
Items
3
Classes
224316
Labels
SK-LARGE
SK-LARGE
SK-LARGE
Human Pose Estimation
Keypoint Skeleton
Microsoft COCO
1491
Items
2
Classes
1491
Labels
CloudCast
CloudCast
A large-scale dataset and baseline for forecasting clouds
Remote Sensing
3D Point Cloud
Aarhus University
70080
Items
11
Classes
70080
Labels
Learning to see in the dark
Learning to see in the dark
Short-exposure night-time images paired with long-exposure images
Computational Photography
Items
Classes
Labels
TEyeD
TEyeD
world's largest unified public data set of eye images
Medical Images
University Tübingen, Germany
20000000
Items
256
Classes
20000000
Labels
DAVIS 2016
DAVIS 2016
Video object segmentation dataset
Video Object Segmentation
Instance Segmentation
Items
Classes
Labels
Distorted Document Images dataset (DDI-100)
Distorted Document Images dataset (DDI-100)
Dataset for Text Detection and Recognition
Text Detection
Bounding Boxes
30000
Items
4
Classes
99870
Labels
MonoPerfCap Dataset
MonoPerfCap Dataset
Dataset for monocular 3D human performance capture
Human Pose Estimation
Keypoint Skeleton
Max Planck Institute for Informatics, Saarland Informatics Campus
40000
Items
20
Classes
40000
Labels
PACS
PACS
Photo-Art-Cartoon-Sketch
Image Classification
Classification Tags
10000
Items
28
Classes
10000
Labels
PedX Dataset
PedX Dataset
Large-scale multi-modal collection of pedestrians at complex urban intersections
Object Detection
Bounding Boxes
PedX
Items
Classes
Labels
MRNet Dataset
MRNet Dataset
A Knee MRI Dataset And Competition
Medical Images
Semantic Segmentation
MRNet
Items
Classes
Labels
Covid19 Challenge Dataset
Covid19 Challenge Dataset
Covid19 Challenge Dataset
Medical Images
Semantic Segmentation
Items
Classes
Labels
Comic2k
Comic2k
Cross-domain object detection dataset
Object Detection
Bounding Boxes
The University of Tokyo, Japan
2000
Items
6
Classes
2000
Labels

Frequently Asked Questions

Looking for more materials to build trustworthy AI? Discover our resources page, packed with free guides, webinars, and V7 product updates.
Where to find machine learning datasets?

One of the best places to look for quality open source datasets is our own repository. You can use advanced filtering options and the search box to look for very specific datasets.

For example, if you’re only interested in a specific licence, such as public domain datasets, make sure to select the CC-0 option in the licence filter.

You can combine various filtering options to narrow down your search.

If you haven’t found what you’re looking for among the 500+ open datasets we’ve catalogued here, don’t despair—there are other places you may want to visit.

Start with these articles:

Each of them comes with detailed descriptions and links to online datasets for various purposes.

Can I use all public datasets on the V7 platform?

Yes, all sample datasets in our repository can be imported into V7.

What are the benefits of using open datasets?

Open datasets offer a number of benefits for computer vision projects. Firstly, they allow for easier collaboration between researchers. When data is openly available, researchers can more easily share and build upon each other’s work. This helps to accelerate the pace of research and allows for more innovative solutions to be found.

Secondly, open datasets help to ensure that the data used is of high quality. When data is openly available, it is subject to greater scrutiny from the research community. This helps to ensure that any flaws or errors in the data are quickly identified and corrected.

Finally, open datasets allow for replicability of results. When data is openly available, researchers can more easily check and verify each other’s results. This helps to build confidence in the findings of a study and allows for more reliable conclusions to be drawn.

Are all datasets on the list free?

Yes, all the online datasets in our repository are free to use. The only limitations may involve the scope of usage or requirements to attribute the dataset to its source. You can easily see what licence a given open dataset falls under on each dataset’s dedicated page.

What kinds of free datasets can I find here?

Our repository of open image datasets consists of free public datasets for computer vision projects. For your convenience we’ve divided them into several categories, e.g.:

Computer vision task types

Use cases

In fact, you can use advanced filtering options to browse our open image datasets by tasks, annotation types, use cases, or licence. Additionally, you can look for interesting datasets by typing a keyword or sorting the results alphabetically, by popularity or the number of images in the dataset.

What are open datasets?

An open dataset for machine learning is a dataset that is freely available for anyone. You can use them as datasets for projects to train and test your machine learning models.

Ready to get started?
Try our trial or talk to one of our experts.