We present a deployed system for on-orbit wildfire detection aboard a nine-satellite commercial thermal infrared constellation, operating under demanding joint constraints: sub-megabyte model footprint, sub-150 ms per-batch TensorRT FP16 inference on an NVIDIA Jetson Xavier NX, and an end-to-end alert pipeline targeting under 10 minutes from satellite overpass to fire event communication. The system operates on uncalibrated mid-wave infrared (MWIR) single-band imagery at 200 m ground sampling distance, where fires frequently appear as sub-pixel or single-pixel thermal anomalies under extreme class imbalance -- challenges not addressed by the contextual thermal-thresholding pipelines (MODIS, VIIRS) that currently dominate operational fire monitoring. We present an empirical study of lightweight dense representation learning for this regime using a proprietary nine-satellite MWIR dataset. We compare dense masked autoencoding (DenseMAE) and a hybrid DenseMAE+EMA (exponential moving average) distillation variant, and evaluate representations via linear probing and full-distribution pixel-level average precision (AP) under extreme class imbalance. DenseMAE pretraining enables compact downstream models on the latency-accuracy Pareto frontier: our fastest SSL-pretrained model achieves 0.640 test AP and 0.69 event-level Fire-F1 with 65.34 ms latency per batch and a 0.52 MB engine, without pruning or compression. The best configuration reaches 0.699 AP and 0.744 Fire-F1 below 1 MB, outperforming a supervised baseline (0.650 AP) under comparable constraints.

翻译：我们提出了一套已部署的在轨野火检测系统，运行于由九颗商业热红外卫星组成的星座之上，同时满足严苛的联合约束条件：模型内存占用低于1兆字节、在NVIDIA Jetson Xavier NX平台上以TensorRT FP16格式进行推理时每批次时间低于150毫秒，以及从卫星过境到火灾事件通信的端到端警报流程控制在10分钟以内。该系统以200米地面采样距离处理未校准的中波红外（MWIR）单波段图像。在该图像中，火灾常表现为亚像素或单像素热异常，且面临极端类别不平衡问题——这些挑战是目前主导业务化火灾监测的基于上下文热阈值处理流程（MODIS、VIIRS）所未能解决的。我们利用专用的九颗卫星MWIR数据集，针对该场景下的轻量级密集表示学习进行了实证研究。我们比较了密集掩码自编码（DenseMAE）与混合型DenseMAE+EMA（指数移动平均）蒸馏变体两种方法，并通过线性探测以及极端类别不平衡下的全分布像素级平均精度（AP）对表示质量进行评估。DenseMAE预训练使得紧凑型下游模型能够处于延迟-精度帕累托前沿：我们最快的自监督学习预训练模型实现了0.640测试AP与0.69事件级Fire-F1分数，每批次延迟为65.34毫秒，引擎大小为0.52 MB，且无需剪枝或压缩。在低于1 MB条件下，最佳配置达到了0.699 AP与0.744 Fire-F1分数，在可比约束下优于有监督基线模型（0.650 AP）。