Niantic, Inc.

The team at Niantic, Inc. is using Meta Llama models to generate creative behaviors in real time as players interact with creatures called "Dots" in Peridot. Using Llama 2, they transform these adorable creatures into responsive AR pets that exhibit the unpredictable nature of real animals. More on how Niantic is using Llama ➡️ https://go.fb.me/us1a0q Using the Niantic Lightship ARDK, the Dots recognize physical objects like flowers, food, and pets — these observations are then processed by a custom version of Llama 2, which considers each Dot’s unique traits to determine how they might react to their discoveries.

Map-free Relocalisation @ ECCV 2024: Call for papers and challenge announcement ($6000 prizes) Niantic, Inc. organizes the first Map-free Relocalisation workshop and challenge, held in conjunction with ECCV 2024 in Milano later this year. Page: https://lnkd.in/e7S3gfVS Map-free Relocalisation (Arnold et al, ECCV 2022) is the problem of estimating a scale-metric relative pose between two RGB images, a reference and a query. Map-free Relocalisation enables instant AR everywhere, without having to build visual maps upfront. The task comes with a substantial dataset that includes challenging image pairs with little or no visual overlap. Reasoning about object shapes is necessary to solve some of the cases. Methods need to estimate a metric scene scale, and provide a robust confidence associated with their pose predictions. We invite the submission of extended abstracts and full papers on the following and related topics: visual relocalization, feature-matching, pose regression (absolute and relative), depth estimation (monocular and multi-frame), scale estimation, confidence and uncertainty, structure-from-motion Paper submission deadline: TBA (~end of July) The challenge features two flavors of Map-Free Relocalisation: Single frame track: The reference is a single RGB image. The query is a single RGB image. Multi-frame track: The reference is a single RGB image. The query is a short burst of frames with tracking poses. Challenge submission deadline: ~2nd August, 2024 $3000 in prizes will be awarded to the top submissions in each of the two tracks. Niantic is also seeking partners from the growing community to co-fund and co-judge the prizes. You can sign up for updates regarding the challenge and workshop here: https://lnkd.in/e3vcSZQK Workshop organizers: Áron Monszpart, Eric Brachmann, Filipe Gaspar, Guillermo Garcia-Hernando, Axel Barroso Laguna, Daniyar Turmukhambetov, Gabriel Brostow, Victor Adrian Prisacariu

Map-free Relocalisation @ ECCV 2024: Call for papers and challenge announcement ($6000 prizes) Niantic, Inc. organizes the first Map-free Relocalisation workshop and challenge, held in conjunction with ECCV 2024 in Milano later this year. Page: https://lnkd.in/e7S3gfVS Map-free Relocalisation (Arnold et al, ECCV 2022) is the problem of estimating a scale-metric relative pose between two RGB images, a reference and a query. Map-free Relocalisation enables instant AR everywhere, without having to build visual maps upfront. The task comes with a substantial dataset that includes challenging image pairs with little or no visual overlap. Reasoning about object shapes is necessary to solve some of the cases. Methods need to estimate a metric scene scale, and provide a robust confidence associated with their pose predictions. We invite the submission of extended abstracts and full papers on the following and related topics: visual relocalization, feature-matching, pose regression (absolute and relative), depth estimation (monocular and multi-frame), scale estimation, confidence and uncertainty, structure-from-motion Paper submission deadline: TBA (~end of July) The challenge features two flavors of Map-Free Relocalisation: Single frame track: The reference is a single RGB image. The query is a single RGB image. Multi-frame track: The reference is a single RGB image. The query is a short burst of frames with tracking poses. Challenge submission deadline: ~2nd August, 2024 $3000 in prizes will be awarded to the top submissions in each of the two tracks. Niantic is also seeking partners from the growing community to co-fund and co-judge the prizes. You can sign up for updates regarding the challenge and workshop here: https://lnkd.in/e3vcSZQK Workshop organizers: Áron Monszpart, Eric Brachmann, Filipe Gaspar, Guillermo Garcia-Hernando, Axel Barroso Laguna, Daniyar Turmukhambetov, Gabriel Brostow, Victor Adrian Prisacariu

Peridot launched a Mixed Reality experience this morning on Quest 3! Check out Hello, Dot in the Meta App Lab Can't wait to build what's next 🔥 https://lnkd.in/gy73__h5

We won the Niantic, Inc. #AI hackathon with our project, Splatley Kubrick! Splatley Kubrick can turn your favorite places into a fantasy sandbox using #GenAI. We spent a week on this project and produced some cool results; check them out! Our team: Zawar Qureshi, Mohamed Sayed, Jamie Watson, Saki Shinoda, Jakub Powierza, and myself.

We won the Niantic, Inc. #AI hackathon with our project, Splatley Kubrick! Splatley Kubrick can turn your favorite places into a fantasy sandbox using #GenAI. We spent a week on this project and produced some cool results; check them out! Our team: Zawar Qureshi, Mohamed Sayed, Jamie Watson, Saki Shinoda, Jakub Powierza, and myself.