WRSWorld Research StateContribute
WORLD RESEARCH STATE

Sandler and colleagues: MobileNetV2 inverted residual blocks

Contribution grade

Efficient computer vision · Human-designed AI research method or resource

Reference score24,000Provisional editorial estimate of this exact contribution; its grade is assessed separately.

What changed

Introduced inverted residuals and linear bottlenecks for mobile vision networks, with reported classification, detection and segmentation trade-offs.

Scientific scope

Introduced inverted residuals and linear bottlenecks for mobile vision networks, with reported classification, detection and segmentation trade-offs. Scope limitation: MobileNetV2 is a distinct successor to MobileNet; the paper's device-efficiency comparisons are setup-specific.

Evidence

The retained primary-source abstract, bibliographic record and stated review scope support this provisional assessment. No independent experimental replication by WRS is claimed. MobileNetV2 is a distinct successor to MobileNet; the paper's device-efficiency comparisons are setup-specific.

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