TIRAuxCloud: A Thermal Infrared Dataset for Day and Night Cloud Detection

Alexis Apostolakis; Vasileios Botsos; Niklas Wölki; Andrea Spichtinger; Nikolaos Ioannis Bountos; Ioannis Papoutsis; Panayiotis Tsanakas

cs.CVFeb 25, 2026

TIRAuxCloud: A Thermal Infrared Dataset for Day and Night Cloud Detection

Authors: Alexis Apostolakis, Vasileios Botsos, Niklas Wölki, Andrea Spichtinger, Nikolaos Ioannis Bountos, Ioannis Papoutsis, Panayiotis Tsanakas

Organizations: National Technical University of Athens, School of Electrical and Computer Engineering · National Observatory of Athens · School of Rural, Surveying and Geoinformatics Engineering, National Technical University of Athens · OroraTech GmbH

Abstract

Clouds are a major obstacle in Earth observation, limiting the usability and reliability of critical remote sensing applications such as fire disaster response, urban heat island monitoring, and snow and ice cover mapping. Therefore, the ability to detect clouds 24/7 is of paramount importance. While visible and near-infrared bands are effective for daytime cloud detection, their dependence on solar illumination makes them unsuitable for nighttime monitoring. In contrast, thermal infrared (TIR) imagery plays a crucial role in detecting clouds at night, when sunlight is absent. Due to their generally lower temperatures, clouds emit distinct thermal signatures that are detectable in TIR bands. Despite this, accurate nighttime cloud detection remains challenging due to limited spectral information and the typically lower spatial resolution of TIR imagery. To address these challenges, we present TIRAuxCloud, a multi-modal dataset centered around thermal spectral data to facilitate cloud segmentation under both daytime and nighttime conditions. The dataset comprises a unique combination of multispectral data (TIR, optical, and near-infrared bands) from Landsat and VIIRS, aligned with auxiliary information layers. Elevation, land cover, meteorological variables, and cloud-free reference images are included to help reduce surface-cloud ambiguity and cloud formation uncertainty. To overcome the scarcity of manual cloud labels, we include a large set of samples with automated cloud masks and a smaller manually annotated subset to further evaluate and improve models. Comprehensive benchmarks are presented to establish performance baselines through supervised and transfer learning, demonstrating the dataset's value in advancing the development of innovative methods for day and night time cloud detection.

Figures & tables

Fig. 1: Global distribution of sampled Landsat scenes used for training, colored by elevation. Each one contains all the cloudy bands mentioned in Section II alongside their clear counterparts and the auxiliary bands. Cloudy bands panel displays Cloudy Radiance B10/B11 mean, while the Clear bands panel shows Clear Radiance B10/B11 mean.

Dataset	Source	Bands	Samples	Size	Resolution	Mask	Notes
		Auxiliary	Number		(Resampled)
38/95-Cloud	Landsat 8/9	VNIR	34,701	384px	30 m	Manual	No TIR
CloudSEN12+	Sentinel 1/2	VNIR, SAR, DEM, LULC	50,249	509px, 2000px	10 m	Manual	No TIR
AllClear	Sentinel 1/2,	VNIR, SAR, DEM,	4,354,652	256px	10 m	Auto	Cloud removal focus;
	Landsat 8/9	LULC, Metadata					small TIR patches
SPARCS	Landsat 8	TIR, VNIR,	80	1000px	30 m	Manual	TIR capable; no auxiliary;

TABLE I: Overview of publicly available remote sensing datasets for cloud detection

Fig. 2: Example bands from a 256×256 Landsat 8 patch.

Fig. 3: Distribution of Landsat and Manually Annotated Landsat patches across Köppen–Geiger climate zones [ 39 ] .

Fig. 4: Temporal distribution of patches across months for VIIRS, Manually Annotated Landsat, and Landsat subsets.

Features	CDnetV2		SegFormer		U-net		SwinCloud
Features	Acc	mIoU	Acc	mIoU	Acc	mIoU	Acc	mIoU
All Bands	0.93	0.86	0.93	0.86	0.93	0.86	0.94	0.89
TIR	0.87	0.75	0.90	0.81	0.90	0.81	0.89	0.79
TIR/Clear	0.87	0.76	0.90	0.81	0.88	0.78	0.86	0.75
TIR/DEM	0.86	0.74	0.90	0.81	0.90	0.80	0.86	0.74
TIR/Meteo	0.89	0.79	0.90	0.81	0.90	0.80	0.90	0.80

TABLE II: Model performance (2-Class Cloud Segmentation) across feature combinations

Features	CDnetV2		SegFormer		U-net		SwinCloud
Features	Acc	mIoU	Acc	mIoU	Acc	mIoU	Acc	mIoU
All Bands	0.91	0.71	0.89	0.66	0.89	0.65	0.90	0.70
TIR	0.83	0.53	0.86	0.56	0.86	0.58	0.80	0.52
TIR/Clear	0.84	0.53	0.86	0.59	0.84	0.55	0.80	0.50
TIR/DEM	0.83	0.51	0.84	0.57	0.85	0.56	0.79	0.49
TIR/Meteo	0.87	0.54	0.88	0.60	0.86	0.55	0.83	0.55

TABLE III: Model performance (3-Class Cloud Segmentation) across feature Combinations

Fig. 5: Examples of inference where auxiliary input contributed to the improvement of accuracy of the model. (a) The model trained only on Cloudy TIR predicts a false positive because the ground in that area exhibited lower than usual thermal values and real clouds were present nearby, causing the region to resemble a cloud. Clear TIR helped resolve this uncertainty. (b) The inclusion of the DEM helps resolve the common infrared based confusion between clouds and high altitude cold terrain (ice/snow), evident in the TIR-only prediction that mistakenly labels part of a mountain as cloud. (c) and (d) The combination of low precipitation and high surface pressure (conditions that suppress deep moist convection and the formation of thick clouds) helps the model more effectively distinguish thin cloud layers from denser cloud cover.

Dataset	Training scheme	U-Net		SegFormer
	FT: Fine Tune	Acc	mIoU	Acc	mIoU
VIIRS	Train	0.79	0.65	0.80	0.66
	Pretrained Inference	0.43	0.22	0.45	0.24
	FT - Frozen encoder	0.77	0.63	0.78	0.63
	FT - Partially Frozen	0.78	0.64	0.80	0.67
MA Landsat	Train	0.87	0.77	0.84	0.72

TABLE IV: Training and fine-tuning results on VIIRS and MA Landsat datasets using Landsat-Main pretraining. Thermal band is used for training.

Remote Sensing Image Segmentation

Remote sensing image segmentation labels land cover and objects in satellite, aerial, radar, and multispectral imagery.

See all 220 papers

Apr 30, 2026

Towards All-Day Perception for Off-Road Driving: A Large-Scale Multispectral Dataset and Comprehensive Benchmark

Off-road nighttime autonomous driving suffers from unreliable visible-light perception, making infrared modality crucial for accurate freespace detection. However, progress remains limited due to the scarcity of annotated infrared off-road datasets and the inter-frame inconsistencies inherent to current single-frame methods. To address these gaps, we present the IRON dataset, which, to our knowledge, is the first large-scale infrared dataset for off-road temporal freespace detection under all-day conditions, with strong support for nighttime perception. The dataset comprises 24,314 densely annotated infrared images with synchronized RGB images in diverse scenes and different light conditions. Building upon this dataset, we propose IRONet, a novel flow-free framework for temporal freespace detection that addresses inter-frame inconsistencies by aggregating historical context via a memory-attention mechanism and a carefully designed mask decoder. On our IRON dataset, IRONet achieves state-of-the-art performance, reaching 82.93%(+1.19%) IoU and 90.66%(+0.71%) F1 score at real-time inference. Remarkably, IRONet also exhibits robust generalization to RGB modalities on ORFD and Rellis datasets. Overall, our work establishes a foundation for reliable all-day off-road autonomous driving and future research in infrared temporal perception. The code and IRON dataset are available at https://github.com/wsnbws/IRON.

Shuo Wang, Jilin Mei, Wenfei Guan +4

Research Center for Intelligent Computing Systems, Institute of Computing Technology, Chinese Academy of Sciences, Beijing, 100190, China. · Beijing Institute of Control Engineering, Beijing, 100194, China.

Autonomous Driving Autonomous Driving Perception

Ask PDF

Sep 26, 2026

Synthetic Thermal Image Generation for Real-Time Animal Detection Under Low-Visibility Conditions

Wildlife-vehicle collisions remain a significant road safety concern, particularly during nighttime and low-visibility conditions when RGB-based perception systems are often unreliable. Thermal imaging offers a promising alternative for detecting animals under poor illumination. However, the limited availability of annotated infrared animal datasets restricts the development of robust deep learning-based detection models. This paper investigates synthetic thermal image generation as a scalable approach for real-time animal detection under low-visibility conditions. A subset of 514 annotated visible-spectrum animal images from the NTLNP dataset is translated into synthetic thermal representations using CycleGAN-Turbo, while a limited real thermal dataset of 60 images is expanded through thermal-focused augmentation. Multiple object detection architectures, including YOLOv8, YOLOv9, YOLOv10, and RT-DETR, are trained independently on synthetic and real thermal datasets and evaluated using precision, recall, [email protected], [email protected]:0.95, model size, and inference latency. Experimental results show that synthetic thermal images provide competitive detection performance, with RT-DETR achieving the highest synthetic-data [email protected] of 0.9613. Models trained on augmented real thermal data achieve the strongest overall performance, with YOLOv10s obtaining 0.9879 [email protected] and 0.9571 [email protected]:0.95. Computational analysis further indicates that lightweight YOLO variants provide favorable inference latency, supporting their potential for real-time deployment. These findings demonstrate that synthetic thermal imagery can reduce dependence on scarce infrared datasets and support the development of efficient animal detection systems for future vehicle-mounted wildlife collision mitigation applications.

James Momoh, Khandaker Mamun Ahmed

The Beacom College of Computer and Cyber Sciences, Dakota State University

Infrared Object Detection Thermal Imaging Synthetic Data Augmentation Real-Time Object Detection Wildlife Monitoring

Ask PDF

Jul 7, 2026

MonoIR-RS: Infrared Remote Sensing Vision-Language Learning with CLIP and VLM Adaptation

Infrared remote-sensing imagery captures intensity structure, object-background contrast, and illumination-invariant cues often invisible in RGB imagery. Yet, most remote-sensing vision-language resources and models focus on visible-band semantics, leaving infrared vision-language understanding underexplored. We introduce MonoIR-RS, a large-scale infrared remote-sensing vision-language dataset and benchmark that couples IR-aware data construction with CLIP-style contrastive adaptation and VLM instruction tuning. Built from the same source pool and split as FusionRS, MonoIR-RS retains the infrared image as the model-facing modality, yielding 600,000 synthesized infrared images and 59,032 retained IR-aware caption records. The model experiments use this retained language-supervision subset, whose captions rewrite supervision around grayscale structure and infrared-style contrast instead of RGB appearance. We show that the synthesized infrared imagery is markedly closer to real thermal imagery than a grayscale conversion on the AVIID benchmark. We fine-tune five CLIP backbones and six VLM backbones, and calibrate them against zero-shot behavior: IR-aware adaptation lifts CLIP mean recall by up to 12.8 points and drives VLM captioning IR-cue coverage to 100% while reducing residual RGB-color leakage to near zero. By isolating the infrared modality from RGB-IR dual-modal learning, MonoIR-RS offers a controlled, reproducible testbed for aligning infrared remote-sensing evidence with language.

Jiaju Han, Ma Yaqi, Yahui Chai +8

China University of Petroleum-Beijing at Karamay, Karamay, Xinjiang, China · Guizhou University, Guiyang, China · Shenzhen Research Institute of Big Data, Shenzhen, China +1

Remote Sensing Image Understanding VLM Adaptation Remote Sensing Multimodal Contrastive Learning Remote Sensing VLMs

Ask PDF

Abstract

Remote Sensing Image Segmentation

Explore similar work

Towards All-Day Perception for Off-Road Driving: A Large-Scale Multispectral Dataset and Comprehensive Benchmark

Synthetic Thermal Image Generation for Real-Time Animal Detection Under Low-Visibility Conditions

MonoIR-RS: Infrared Remote Sensing Vision-Language Learning with CLIP and VLM Adaptation