Deep learning-based detection of construction waste in complex scenarios with an improved lightweight algorithm
YiZhong Yang, YeXue Li, ShengWei Li, MaoHu Tao, Xuzhi Chen · 2026
Read Deep learning-based detection of construction waste in complex scenarios with an improved lightweight algorithm by YiZhong Yang, YeXue Li, ShengWei Li, MaoHu Tao, Xuzhi Chen (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-0356914.txt.
Description
YiZhong Yang leads a team of five teaching a camera system to sort construction waste on sites, in PLOS ONE in 2026. Rubble heaps mix categories that look alike, hide under each other, and bend out of shape, while site computers are weak. Their algorithm trims a detection network for lightness. It adds spatial attention so the model notices where things are. It also redesigns the detection head, aligning what the model sees with what it is asked to find.
On the team's ten category dataset, the system beat its parent network on detection quality and precision while shedding about a fifth of its parameters. The model weighs four megabytes, small enough for cheap site hardware to run in real time. Tests on public datasets backed its ability to generalize beyond its home set. For an industry drowning in debris, automated sorting at the source starts with eyes this cheap.
Catalog record
| CREATOR | YiZhong Yang, YeXue Li, ShengWei Li, MaoHu Tao, Xuzhi Chen |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 97324 |
| LICENSE | CC_BY |
| RIGHTS | CC BY |
| SOURCE | PLOS |