Benchmark RadarAI BENCHMARK PROFILE
EgoAfford
EgoAfford is a benchmark for egocentric referring segmentation with task-oriented affordance grounding, comprising 15.5k images and 102 real images. It includes EgoLens, a 3B MLLM reference model.
- Released
- 2026-08-05
- Readiness
- Inspectable
- Primary field
- General AI
Why it matters
It addresses the need for connecting perception and planning in tabletop tasks, but the lack of public artifacts and scoring details limits its immediate utility.
Motivation
Part-level affordance grounding has advanced the localization of functional object regions associated with elemental actions.
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.