AI Paper Atlas

Research map / Detection and segmentation / CNN features

NAPA: Natively Multimodal Autoregressive Perception Architecture

Phuc Le Khac Hong, Yasser Dahou, Sanath Narayan, Ankit Singh, Wamiq Para, Ngoc Dung Huynh, Sofian Chaybouti, Hakim Hacid

ECCV 2026 · Detection and segmentation › CNN features · approach: Plain and high-resolution ViTs

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use Muon instead of AdamW optimizer (ablation)Falcon Perception (detection) · SaCo-DetMacro F1: 49.0 -> 53.8Improved
use Muon instead of AdamW optimizer (ablation)Falcon Perception (detection) · PBench-DetMacro F1: 56.4 -> 57.7Improved
add Gram loss feature regularization (ablation)Falcon Perception (segmentation) · PBench-SegMacro F1: 52.7 -> 53.8Improved
add Gram loss feature regularization (ablation)Falcon Perception (segmentation) · SaCo-SegMacro F1: 51.1 -> 52.6Improved
mask attention between queries instead of full autoregressive attention (ablation)Falcon Perception (detection) · PBench-DetMacro F1: 53.2 -> 54.2Improved
mask attention between queries instead of full autoregressive attention (ablation)Falcon Perception (detection) · SaCo-DetMacro F1: 53.1 -> 53.3No clear effect
use raster ordering of instances instead of random ordering (ablation)Falcon Perception (detection) · SaCo-DetMacro F1: 46.3 -> 56.2Improved
use raster ordering of instances instead of random ordering (ablation)Falcon Perception (detection) · PBench-DetMacro F1: 52.2 -> 59.3Improved
use multi-teacher distillation initialization instead of random initialization (ablation)Falcon Perception (300M, detection) · SA-CopmF1 and MCCImproved
increase input resolution from 448^2 to 1024^2 (ablation)Falcon Perception · PBench Densemicro-F1 (%): 3.9% -> 61.0%Improved
increase input resolution from 448^2 to 1024^2 (ablation)Falcon Perception · SA-CoIL_MCC: 0.34 at 448^2; MCC grows 1.6xImproved
sample k=8 predictions and take the best (Pass@8) instead of deterministic decoding (ablation)Falcon Perception · SA-Coaverage cgF1: 34.7 -> 54.3Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy