### About this Dataset

Defined.ai is now offering a comprehensive short form videos dataset. This meticulously collected and curated collection was developed with the full consent of each user, in strict adherence to each platform's terms and conditions. This dataset provides an invaluable resource for researchers, data scientists, and businesses interested in multi-model training. Each video comes with classification annotations for gender, age, ethnicity, and more.

#### License Information

This dataset is covered by our standard [Data license agreement](https://www.defined.ai/dataset/data-licence-agreement). The license agreement is perpetual and allows for the commercialization of all models built on the data.


Tell us about yourself, and get access to the dataset samples

Download sample of short videos dataset via this [link](https://prdstrapimediastorage.blob.core.windows.net/prdstrapimediastorage/assets/sample_shortform_video_ed47f131b5.zip?updated_at=2023-06-02T18:17:19.089Z)

Download Sample

Image and Video

Sensitive Images and Videos

Identity Verification Dataset

Retail Images Dataset

Speech

Online Meeting Recordings

You might also be interested in:

Short Form Videos Dataset

Defined.ai has heard our clients' need for more diverse data, and for ethically sourced live data with full consent. If you are looking for real, online meeting recordings with full audio and video, along with and timestamped transcription, look no further.

Our collection of online meeting recordings captures authentic Zoom meetings that cover topics like engineering, marketing, product, facilities, and much more. This dataset was conceived from the get-go to have privacy and consent in mind, with all participants using nicknames and giving explicit consent for their voices to be used for AI modeling purposes. Meetings are held in English by a mix of native and non-native speakers or in Portuguese by native speakers.

These online meeting recording datasets are going to be available by the end of 2023. You can already contact us to get samples and more information about these datasets!

Zoom-recorded meetings covering topics like engineering, marketing, facilities, and more

### About this dataset

This retail images dataset focuses on the brick-and-mortar retail domain, primarily from locations in Brazil, and contains almost 8 million images of assorted items on shelves, storefronts, close-ups of price tags, and more.

With this data, the builders of AI will be able to train new models to cover copious use cases. The retail images dataset was sourced in accordance with Defined.ai's strict ethical standards, ensuring that all data was obtained with the informed consent of those involved and that privacy rights were respected. 

We believe that ethical sourcing of AI is of utmost importance, and we take that responsibility seriously. We are confident that our clients will appreciate the care and attention we put into curating this data.

The dataset contains:
- photos of convenience store
- photos of checkout stands in a convenience store
- photos of shelving facades in a convenience store
- photos of price tags in a convenience store

#### License Information

This dataset is covered by our standard [Data license agreement](https://www.defined.ai/dataset/data-license-agreement). The license agreement is perpetual and allows for the commercialization of all models built on the data.

### Samples
#### Preview

![retail_price_tags_preview.jpg](https://prdstrapimediastorage.blob.core.windows.net/prdstrapimediastorage/assets/retail_price_tags_preview_4217b9b567.jpg)
![retail_price_tags_preview_2.jpg](https://prdstrapimediastorage.blob.core.windows.net/prdstrapimediastorage/assets/retail_price_tags_preview_2_b877cfc852.jpg)

8 million images of assorted items on shelves, storefronts, and more

### About this dataset

Open-source datasets often fall short of providing quality and diversity, specifically in the field of facial recognition. Our carefully curated datasets, composed of diverse selfies and paired ID photos, elevate the standard, offering robust resources for superior facial recognition analysis and model development.

- 366,419 images
- 40,717 contributors
- 7 ethnic groups: South Asian, Native American, Middle Eastern, Hispanic, East Asian, Caucasian, and African
- 2 IDs & 8 Selfies per person
- 1 – 4 Skintones
- 50/50 Gender Distribution
- 18 – 65 Age Groups
- Different camera & lighting conditions

From organic to paid acquisition, we actively recruit contributors from a range of sources to create a demographically diverse crowd. To source our data ethically and transparently, all contributors are paid fairly and informed about how their data is used. Privacy is also of utmost concern as we're GDPR-compliant and ISO-27001 certified, and contributors agree to non-disclosure agreements and master service agreements prior to collection. Read more about our safe, secure, and ethical collection processes here.

#### License Information

This dataset is covered by our standard [Data License Agreement](https://www.defined.ai/dataset/data-license-agreement). The license agreement is perpetual and allows for the commercialization of all models built on the data.

### Samples
![UG10052_8.jpg](https://prdstrapimediastorage.blob.core.windows.net/prdstrapimediastorage/assets/UG_10052_8_cc7e68d644.jpg)
![UG10052_9.jpg](https://prdstrapimediastorage.blob.core.windows.net/prdstrapimediastorage/assets/UG_10052_9_116b06abee.jpg)

![UP11652_3.jpg](https://prdstrapimediastorage.blob.core.windows.net/prdstrapimediastorage/assets/UP_11652_3_f57e48a38b.jpg)
![UP11652_1.jpg](https://prdstrapimediastorage.blob.core.windows.net/prdstrapimediastorage/assets/UP_11652_1_70abf5bb66.jpg)

Diverse facial image data for ID verification models

### About this dataset

We are pleased to offer a selection of high-quality AI training data for teams working in content moderation and user trust and safety. Our datasets are designed to train computer vision models to identify and flag inappropriate content automatically, such as nudity, without the need for a human-powered moderation team to be exposed to it.

Our data was ethically collected with fair compensation and full consent. It also includes the redaction of all personally identifiable information (PII). Our commitment to ethical data collection means that you can use our training data with confidence, knowing that it has been collected with respect for privacy and human dignity.

By using our training data, you can help ensure that your digital platform or application provides a safe and welcoming environment for all users. Whether you are just starting out or you are an experienced data scientist, our content moderation datasets can help you achieve your goals. Contact us to get samples and more information about these datasets.

The dataset contains male and female selfie images and videos captured with front-facing camera.

#### License Information

This dataset is covered by our standard [Data license agreement](https://www.defined.ai/dataset/data-license-agreement). The license agreement is perpetual and allows for the commercialization of all models built on the data.

1500 individuals and over 490,000 images, fully indexable by ethnicity, gender, and more

TikTok videos collected and curated with the full consent of each user