Optimizing multi-label image annotation: a hybrid CNN-Transformer deep learning approach
Chitrapriya Ningthoujam, Tejbanta Singh Chingtham, Dhruba Ningombam · 2026
Read Optimizing multi-label image annotation: a hybrid CNN-Transformer deep learning approach by Chitrapriya Ningthoujam, Tejbanta Singh Chingtham, Dhruba Ningombam (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/pmc-13408854.txt.
Description
Chitrapriya Ningthoujam and co-authors built a hybrid AI for tagging photos with many labels in this 2026 paper. A photo can hold a dog, a beach, and a sunset at once, and tag sets are messy, lopsided, and noisy. Their model, ReSwinNet, pairs a classic image network with a transformer that reads across the whole scene.
A bridge step feeds the first network's fine detail into the transformer without losing it to a coarse grid. A pruning step drops labels that appear too rarely or unevenly to learn. On two benchmark sets the model beat most or all rivals, with the top recall in one test. The gains grew under the cleaned label sets the pruning produced.
Catalog record
| CREATOR | Chitrapriya Ningthoujam, Tejbanta Singh Chingtham, Dhruba Ningombam |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 62 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |