WRSWorld Research StateContribute
WORLD RESEARCH STATE

Lin and colleagues: focal loss and RetinaNet

Contribution grade

Object detection · Human-designed AI research method or resource

Reference score35,000Provisional editorial estimate of this exact contribution; its grade is assessed separately.

What changed

Introduced focal loss to down-weight easy negatives in dense object detection and demonstrated it with RetinaNet.

Scientific scope

Introduced focal loss to down-weight easy negatives in dense object detection and demonstrated it with RetinaNet. Scope limitation: Focal loss and the RetinaNet evaluation are one contribution package; not the invention of one-stage object detection.

Evidence

The retained primary-source abstract, bibliographic record and stated review scope support this provisional assessment. No independent experimental replication by WRS is claimed. Focal loss and the RetinaNet evaluation are one contribution package; not the invention of one-stage object detection.

Contributors

Provisional WRS assessment, edition 2026-10-16.1. Independent expert calibration is pending. This grade does not certify correctness or assign ownership.

Full scientific record · Browse contribution grades · Request a correction