FoundYou paper accepted at ECCV 2026: one photo tracks and segments the same object across thousands of images
gabriberton · x · 2026-09-03
The team behind FoundYou: A Unified Model for Personalized Segmentation and Retrieval announced the paper's acceptance at ECCV 2026.
- Core capability: snap a photo of an item (e.g., a t-shirt) and the model can search through thousands of images, find that exact item, and segment it every time it appears.
- Positioning: unifies personalized segmentation and retrieval into a single model, going beyond generic segmentation models like Segment Anything 2.
More from Multimodal
- Playable Mars rover built with Spline V2's new AI agent, iteratively prompted — dunkhippo33 · 2026-09-03
- Long-time Midjourney user shares first ComfyUI img2img + LoRA workflow on a 24GB M4 Pro MacBook — the_ground_state · 2026-09-03
- AST-based ComfyUI prompt parser adds inline negative routing and weighted wildcards — overlord_sid85 · 2026-09-03
- Moondream's Photon 2.1 adds streaming ASR, beats vLLM with up to 3.1x throughput — xeophon · 2026-09-03
- Atlas can turn a single reference image into novel views from any camera position and angle — XRarchitect · 2026-09-03
- Atlas: text-to-image is just single-frame generation under one unified formulation — kdexd · 2026-09-03