DOLD-Net: Dense occluded livestock detection via global-local feature collaboration
Kaida Jia, Yang Chai, Baozhou Chen, Zunhai Gao · 2026
Read DOLD-Net: Dense occluded livestock detection via global-local feature collaboration by Kaida Jia, Yang Chai, Baozhou Chen, Zunhai 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/plos-pone-0356771.txt.
Description
Kaida Jia, Yang Chai, Baozhou Chen, and Zunhai Gao present DOLD-Net, a vision model built for one of farming's hardest counting problems. The problem is animals packed body to body, hiding each other. Their 2026 paper assembles a two-branch backbone. It reads both fine edges and overall herd layout. A mechanism pushes context into hidden regions. Other modules sharpen blurred boundaries between overlapping animals.
Across four public livestock datasets, the model posted strong accuracy scores. It stayed efficient enough for farm deployment. It beat the usual trade-off between precision and speed. Reliable animal detection underpins counting, health checks, and welfare monitoring in smart barns. Occlusion is precisely where simpler detectors fall apart. The authors have released their code publicly.
Catalog record
| CREATOR | Kaida Jia, Yang Chai, Baozhou Chen, Zunhai Gao |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 56263 |
| LICENSE | CC_BY |
| RIGHTS | CC BY |
| SOURCE | PLOS |