Lightweight detection network based on bidirectional weighted feature fusion with small target enhancement for USVs

Yong Li, Dehang Lian, Jialong Du, Chunning Bu, Dongxu Gao · 2026

Read Lightweight detection network based on bidirectional weighted feature fusion with small target enhancement for USVs by Yong Li, Dehang Lian, Jialong Du, Chunning Bu, Dongxu Gao (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/PMC13562531.txt.

Description

Yong Li and four colleagues build a lean vision network that helps robot boats spot small floating rubbish on water. Uncrewed surface vessels collect drifting waste, yet waves, glare, and shore shadows confuse ordinary detectors, and floating items are often tiny in the frame. The 2026 design, named SEBR-YOLOv8n, reworks a compact detector with a two-way weighted feature pyramid that boosts small targets. A transformer-based decoder and a refined box-matching loss sharpen the results further. The whole model holds only five million parameters.

On a standard floating-waste image set, the network reached detection scores of about 91 and 50 percent, beating its base model by roughly six and four points. A second dataset confirmed the gain, and images with synthetic extra samples scored higher still. The authors argue the design gives cleanup vessels sharp eyes without heavy computing hardware.

Catalog record

CREATORYong Li, Dehang Lian, Jialong Du, Chunning Bu, Dongxu Gao
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE54383
LICENSEhttps://creativecommons.org/licenses/by/4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/article/MED/PMC13562531

Browse the full library · About the Tower of Babel