arXiv：cs.LG· Spandan Ghose Chowdhury·· 4 小时前AI 评分39
用卫星验证闭合航迹云规避环路：一个小型扩散模型检测航迹云的研究
Closing the Loop on Contrail Avoidance with Satellite Verification
AI 导读
研究人员训练了一个8.4M参数、单GPU训练的小型扩散模型来检测航迹云，PR-AUC达0.476，高于DeepLabV3+基线的0.414和适配版MedSegDiff的0.119。将CNN输入分辨率翻倍后基线达到0.499，与扩散模型持平（p=0.07）。研究表明简单翻转旋转等增强使准确率提升一倍以上，而用航迹云形状预训练反而有害——模型精确率崩塌至1%，且基于召回率的指标仍误判其表现优异。
正文
Abstract:Contrails are the thin ice clouds that aircraft leave behind. They cause a large share of aviation's warming, and rerouting the few flights that produce them could avoid much of it. However, an avoided contrail only counts if a satellite can confirm that it never formed, and this check is hard: contrails are one to two pixels wide, cover only 0.18% of pixels, and look very similar to natural cirrus. We build a small diffusion model (8.4M parameters, trained on one GPU) that detects them, and we run a controlled study to find out which components matter. The model reaches 0.476 PR-AUC, compared with 0.414 for a DeepLabV3+ baseline and 0.119 for an adapted MedSegDiff. Doubling the input resolution of the CNN brings it to parity (0.499, p=0.07). Three lessons apply beyond contrails. First, check the input resolution before designing a new architecture. Second, simple flips and rotations more than double accuracy and matter more than any architectural choice we measured. Third, pretraining the model on contrail shapes is harmful: the model learns that thin strokes appear everywhere and paints them onto empty scenes. Precision collapses to 1% while recall-based metrics still rate the degraded model as excellent, and no threshold or guidance heuristic repairs this failure.
| Comments: | Accepted into Tackling Climate Change with Machine Learning: workshop at NeurIPS 2026 |
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG) |
| Cite as: | arXiv:2610.09363 [cs.CV] |
| (or arXiv:2610.09363v1 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2610.09363 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Spandan Ghose Chowdhury [view email]
[v1]
Wed, 7 Oct 2026 03:18:29 UTC (2,089 KB)
来源：arXiv：cs.LG · arxiv.org