跳到正文
arXiv cs.AI· Chanhwi Jeong, Dongmin You, Seungwook Kim·· 1 小时前AI 评分32

TOGBench:评测并提升航拍图像微小目标生成保真度

TOGBench: Benchmarking and Improving Tiny-Object Generation Fidelity in Aerial Imagery

AI 导读

研究者提出 TOGBench,一个面向航拍图像中微小目标(如车辆、船只、飞机)生成保真度的定量、对象级基准。现有生成模型对小目标的渲染明显弱于较大目标,而以往评估多强调提示词遵循或人类偏好,忽视了这一问题。论文还提出两种训练与推理阶段的简单技术,提升 layout-to-image 生成器渲染小物体的表现,该工作已被 WACV 2027 Round 2 接收。

正文 · 原文

View PDF HTML (experimental)

Abstract:Advances in deep generative models now enable the synthesis of realistic imagery. However, aerial imagery is dominated by tiny objects, such as vehicles, ships, and aircraft, which occupy only a few pixels owing to the overhead, wide-area nature of aerial imaging. Existing generative models struggle substantially more with tiny objects than with larger ones. This limitation has been overlooked, as prior work on generative evaluation emphasizes prompt following or human preference rather than whether small objects are rendered faithfully. To address this gap, we introduce TOGBench, a quantitative, object-level benchmark for evaluating tiny-object generation fidelity in aerial imagery. We further propose two simple train-/inference-time techniques that improve the ability of layout-to-image generators to render small objects, raising their performance on TOGBench.
Comments: Accepted to WACV 2027 Round 2. 20 pages, 7 figures
Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
Cite as: arXiv:2611.23840 [cs.CV]
  (or arXiv:2611.23840v1 [cs.CV] for this version)
  https://doi.org/10.48550/arXiv.2611.23840

arXiv-issued DOI via DataCite (pending registration)

Submission history

From: Seungwook Kim [view email]
[v1] Wed, 18 Nov 2026 04:45:57 UTC (15,847 KB)

来源:arXiv cs.AI · arxiv.org