Payments now enabled in Trove
Trove is a marketplace app that allows people to contribute photos to AI projects that developers can then use to train machine learning models. They’ve tested early versions of the experience with thousands of users, inspired by the desire to create a place where people can get rewarded for their photos, the Trove team has enabled payments for accepted photo contributions.
Now, users can set up or connect a PayPal account, scroll through available projects and their associated photo offer prices, and submit new or existing photos using the newly released Camera Roll feature. When photos are accepted, users will be notified of payment and can find the reward in their PayPal balance.
Today, users might find AI projects looking for photos on books, magazines, and newspapers, hand-written notes, or numerical digital displays (such as a digital watch or clock). The reward per photo varies by the project, but users might expect to be paid roughly 50 or 60 cents per approved photos.
Feedback loops and quality inputs for research projects are also driving objectives of the marketplace, so the new feature also gives AI researchers and developers the opportunity to provide payment upon approval of photos, and optionally, to set per user photo contribution cap to ensure a variety of submissions.
Responsible AI starts with responsible data collection
Trove was created and imagined by an incubation team who was inspired to bridge their passion for AI advancement with their vision for people-first gig work that promotes transparency and choice. In Trove, they’ve built a marketplace that enables researchers to build machine learning models using responsible data collection. Read the full origin story and inspiration behind Trove in our May 2020 launch blog.
Photo contributors have reported being drawn to the opportunity to contribute to AI and the ideals of a transparent, empowering marketplace. Sebastian, a photo contributor in the earliest versions of Trove, shared, “These little projects are fun to do and it feels like I am doing something bigger than I could ever do on my own.”
“Again and again, we heard from our earliest users that they were excited to be contributing to the march of progress and science, to be part of something bigger than themselves,” shared Christian Lisenberger, one of the Trove project leads. “It is so rewarding to see that we’re able to provide an alternative, responsible way to both advance AI and give people more agency over data and gig work.”
To learn more about Microsoft’s approach to Responsible AI and learn more about Trove, watch Mitra Azizirad’s Microsoft Ignite 2020 talk, available live at 9:55 – 10:40 AM Pacific Time and later, on demand.
New features inspired by community feedback
In addition to being more broadly available, the team has added new key features to the user experience. When the team set out to create Trove, they were inspired to make a people-first solution. They’ve spent the past few months working with customers to understand how to improve their photo contribution or AI project experiences.
Notably, many users expressed excitement for uploading photos from their camera roll, in addition to taking new photos directly from the app. The team is hopeful this will make it easier to bulk upload existing photos and expand the photo variety and volume researchers can include in their computer vision projects.
- Payments for accepted photos Receive monetary compensation for every approved photo that you submit.
- Camera roll upload Upload photos from camera roll or take a new picture via the app
- Community forum Submit feedback via answers.microsoft.com
For the full list of key features and trial details, see the Trove Garage Workbench page.
Try it out
Trove is now available as a web app experience, or for download via the Google Play Store for users in the United States. The team continues to be excited for feedback and encourages users to share their thoughts and ideas.