{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,3,18]],"date-time":"2026-03-18T02:50:14Z","timestamp":1773802214236,"version":"3.50.1"},"reference-count":0,"publisher":"Association for the Advancement of Artificial Intelligence (AAAI)","issue":"16","content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["AAAI"],"abstract":"<jats:p>Neural Radiance Fields (NeRF) have shown remarkable capabilities\nfor photorealistic novel view synthesis. One major deficiency of\nNeRF is that dense inputs are typically required, and the rendering\nquality will drop drastically given sparse inputs. In this paper, we\nhighlight the effectiveness of rendered semantics from dense novel\nviews, and show that rendered semantics can be treated as a more\nrobust form of augmented data than rendered RGB. Our method\nenhances NeRF\u2019s performance by incorporating guidance derived\nfrom the rendered semantics. The rendered semantic guidance encompasses\ntwo levels: the supervision level and the feature level.\nThe supervision-level guidance incorporates a bi-directional verification\nmodule that decides the validity of each rendered semantic\nlabel, while the feature-level guidance integrates a learnable codebook\nthat encodes semantic-aware information, which is queried\nby each point via the attention mechanism to obtain semanticrelevant\npredictions. The overall semantic guidance is embedded\ninto a self-improved pipeline.We also introduce a more challenging\nsparse-input indoor benchmark, where the number of inputs is\nlimited to as few as 6. Experiments demonstrate the effectiveness\nof our method and it exhibits superior performance compared to\nexisting approaches.<\/jats:p>","DOI":"10.1609\/aaai.v40i16.38359","type":"journal-article","created":{"date-parts":[[2026,3,18]],"date-time":"2026-03-18T00:24:57Z","timestamp":1773793497000},"page":"13539-13547","source":"Crossref","is-referenced-by-count":0,"title":["Empowering Sparse-Input Neural Radiance Fields with Dual-Level Semantic Guidance from Dense Novel Views"],"prefix":"10.1609","volume":"40","author":[{"given":"Yingji","family":"Zhong","sequence":"first","affiliation":[]},{"given":"Kaichen","family":"Zhou","sequence":"additional","affiliation":[]},{"given":"Zhihao","family":"Li","sequence":"additional","affiliation":[]},{"given":"Lanqing","family":"Hong","sequence":"additional","affiliation":[]},{"given":"Zhenguo","family":"Li","sequence":"additional","affiliation":[]},{"given":"Dan","family":"Xu","sequence":"additional","affiliation":[]}],"member":"9382","published-online":{"date-parts":[[2026,3,14]]},"container-title":["Proceedings of the AAAI Conference on Artificial Intelligence"],"original-title":[],"link":[{"URL":"https:\/\/ojs.aaai.org\/index.php\/AAAI\/article\/download\/38359\/42321","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/ojs.aaai.org\/index.php\/AAAI\/article\/download\/38359\/42321","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,3,18]],"date-time":"2026-03-18T00:24:58Z","timestamp":1773793498000},"score":1,"resource":{"primary":{"URL":"https:\/\/ojs.aaai.org\/index.php\/AAAI\/article\/view\/38359"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,3,14]]},"references-count":0,"journal-issue":{"issue":"16","published-online":{"date-parts":[[2026,3,17]]}},"URL":"https:\/\/doi.org\/10.1609\/aaai.v40i16.38359","relation":{},"ISSN":["2374-3468","2159-5399"],"issn-type":[{"value":"2374-3468","type":"electronic"},{"value":"2159-5399","type":"print"}],"subject":[],"published":{"date-parts":[[2026,3,14]]}}}