Far-infrared Imaging Research Articles

Multispectral Pedestrian Detection Based on Prior-Saliency Attention and Image Fusion

Detecting pedestrians in varying illumination conditions poses a significant challenge, necessitating the development of innovative solutions. In response to this, we introduce Prior-AttentionNet, a pedestrian detection model featuring a Prior-Attention mechanism. This model leverages the stark contrast between thermal objects and their backgrounds in far-infrared (FIR) images by employing saliency attention derived from FIR images via UNet. However, extracting salient regions of diverse scales from FIR images poses a challenge for saliency attention. To address this, we integrate Simple Linear Iterative Clustering (SLIC) superpixel segmentation, embedding the segmentation feature map as prior knowledge into UNet’s decoding stage for comprehensive end-to-end training and detection. This integration enhances the extraction of focused attention regions, with the synergy of segmentation prior and saliency attention forming the core of Prior-AttentionNet. Moreover, to enrich pedestrian details and contour visibility in low-light conditions, we implement multispectral image fusion. Experimental evaluations were conducted on the KAIST and OTCBVS datasets. Applying Prior-Attention mode to FIR-RGB images significantly improves the delineation and focus on multi-scale pedestrians. Prior-AttentionNet’s general detector demonstrates the capability of detecting pedestrians with minimal computational resources. The ablation studies indicate that the FIR-RGB+ Prior-Attention mode markedly enhances detection robustness over other modes. When compared to conventional multispectral pedestrian detection models, Prior-AttentionNet consistently surpasses them by achieving higher mean average precision and lower miss rates in diverse scenarios, during both day and night.

Effects of working fluid on performance of 4.5 K JT cooler

NASA is considering a “probe” class mission with a far-infrared imaging and spectroscopy space telescope. A critical technology need for this mission is an efficient, high capacity, low-temperature cryocooler with low exported vibrations to enable its cryogenic instruments to achieve exquisite sensitivity and low internal noise to observe far distant faint cosmic sources. This paper discusses the design of a 4.5 K hybrid cooler consisting of a pulse tube precooler and a JT stage. The JT stage uses 3He as its working fluid to achieve a low cooling temperature of 4.5 K. The cooler will build on the JWST MIRI spare flight cooler, using existing TRL 9 component and subsystem hardware technologies. The paper will discuss the cooler configuration, thermodynamic design, predicted performance, and sensitivity to operating conditions. The impact of pressure drops and thermal ineffectiveness in the JT stage recuperators are discussed and compared to those in the MIRI cooler. Finally, test results for MIRI Development Model (DM) compressor with 4He are presented. Test data include compressor volumetric flow rates and power inputs at different pressure ratios and piston stroke lengths. The DM compressor test results and flight compressor test data collected during MIRI spare flight cooler acceptance testing both indicate that the 4.5 K cooler will be able to meet the cooling requirement of 48 mW at 4.5 K and 230 mW at 18 K with a significant margin.

Fall Detection in Low-Illumination Environments From Far-Infrared Images Using Pose Detection and Dynamic Descriptors

Fall Detection in Low-Illumination Environments From Far-Infrared Images Using Pose Detection and Dynamic Descriptors

Detecting a disc bending wave in a barred-spiral galaxy at redshift 4.4

ABSTRACT The recent discovery of barred spiral galaxies in the early Universe (z > 2) poses questions of how these structures form and how they influence galaxy evolution in the early Universe. In this study, we investigate the morphology and kinematics of the far-infrared (FIR) continuum and [C ii] emission in BRI1335-0417 at z ≈ 4.4 from ALMA observations. The variations in position angle and ellipticity of the isophotes show the characteristic signature of a barred galaxy. The bar, $3.3^{+0.2}_{-0.2}$ kpc long in radius and bridging the previously identified two-armed spiral, is evident in both [C ii] and FIR images, driving the galaxy’s rapid evolution by channelling gas towards the nucleus. Fourier analysis of the [C ii] velocity field reveals an unambiguous kinematic m = 2 mode with a line-of-sight velocity amplitude of up to ∼30–40 km s−1; a plausible explanation is the disc’s vertical bending mode triggered by external perturbation, which presumably induced the high star formation rate and the bar/spiral structure. The bar identified in [C ii] and FIR images of the gas-rich disc galaxy (≳ 70 per cent of the total mass within radius R ≈ 2.2 disc scale lengths) suggests a new perspective of early bar formation in high redshift gas-rich galaxies – a gravitationally unstable gas-rich disc creating a star-forming gaseous bar, rather than a stellar bar emerging from a pre-existing stellar disc. This may explain the prevalent bar-like structures seen in FIR images of high-redshift submillimeter galaxies.

Dissecting the morphology of star forming complex S193

ABSTRACT We have studied a star-forming complex S193 using near-infrared (NIR) observations and other archival data covering optical to radio wavelengths. We identified stellar clusters in the complex using the NIR photometric data and estimated the membership and distance of the clusters. Using the mid-infrared (MIR) and far-infrared (FIR) images, the distribution of the dust emission around H ii regions is traced in the complex. The Herschel column density and temperature maps analysis reveal 16 cold dust clumps in the complex. The H α image and 1.4 GHz radio continuum emission map are employed to study the ionized gas distribution and infer the spectral type and the dynamical age of each H ii region/ionized clump in the complex. The 12CO(J = 3 − 2) and 13CO(J = 1 − 0) molecular line data hint at the presence of two velocity components around [−43, −46] and [−47, −50] km s−1, and their spatial distribution reveals two overlapping zones toward the complex. By investigating the immediate surroundings of the central cluster [BDS2003]57 and the pressure calculations, we suggest that the feedback from the massive stars seems responsible for the observed velocity gradient and might have triggered the formation of the central cluster [BDS2003]57.

AFGL 5180 and AFGL 6366S: sites of hub–filament systems at the opposite edges of a filamentary cloud

ABSTRACT We present a multiscale and multiwavelength study to unveil massive star formation (MSF) processes around sites AFGL 5180 and AFGL 6366S, both hosting a Class ii 6.7 GHz methanol maser emission. The radio continuum map at 8.46 GHz reveals a small cluster of radio sources towards AFGL 5180. Signatures of the early stages of MSF in our target sites are spatially seen at the opposite edges of a filamentary cloud (length ∼5 pc), which is observed in the submillimetre dust continuum maps. Using the near-infrared photometric data, the spatial distribution of young stellar objects is found towards the entire filament, primarily clustered at its edges. The getsf utility on the Herschel far-infrared images reveals a hub–filament system (HFS) towards each target site. The analysis of the molecular line data, which benefits from large area coverage (∼1° × 1°), detects two cloud components with a connection in both position and velocity space. This supports the scenario of a cloud–cloud collision (CCC) that occurred ∼1 Myr ago. The filamentary cloud, connecting AFGL 5180 and AFGL 6366S, seems spatially close to an H ii region Sh 2−247 excited by a massive O9.5 star. Based on the knowledge of various pressures exerted by the massive star on its surroundings, the impact of its energetic feedback on the filamentary cloud is found to be insignificant. Overall, our observational outcomes favour the possibility of the CCC scenario driving MSF and the formation of HFSs towards the target sites.

The Photodissociation and Ionization Fronts in M17-SW Localized with FIFI-LS on Board SOFIA

To understand star formation rates, studying feedback mechanisms that regulate star formation is necessary. The radiation emitted by nascent massive stars play a significant role in feedback by photodissociating and ionizing their parental molecular clouds. To gain a detailed picture of the physical processes, we mapped the photodissociation region (PDR) M17-SW in several fine-structure and high-J CO lines with FIFI-LS, the far-infrared imaging spectrometer aboard SOFIA. An analysis of the CO and [O i]146 μm line intensities, combined with the far-infrared intensity, allows us to create a density and UV intensity map using a one-dimensional model. The density map reveals a sudden change in the gas density crossing the PDR. The strengths and limits of the model and the locations of the ionization and photodissociation front of the edge-on PDR are discussed.

Study of an Isolated Dust Structure Nearby the White Dwarf WD0011-399 Using IRIS, AKARI, and WISE Data

During the asymptotic giant branch (AGB) phase, a major fraction of mass is spread by the stars (0.6 - 10 M⊙) in the interstellar medium (ISM) in the form of dust. In the left phase of the post-main sequence evolution in the Hertzsprung-Russell (HR) diagram, these dust are found to be surrounding the white dwarfs (WD). Some small fraction of dust is also formed in the circumstellar shells and cavities around WD which is usually a source of infrared (IR) excess. The formation and evolution of such IR dust structures are the results of high-pressure events such as the Helium shell flash (thermal pulse) and stellar winds, and such structures are crucial in the study of interaction phenomena in ISM. This research project is focused on the study of an isolated dust structure near the white dwarf WD0011-399 located at R.A. (J2000) 00h 13m 47.48s and Dec. (J2000) -39° 370 24.2800 using Improved Reprocessing of the IRAS (IRIS), AKARI and Wide-field Infrared Survey Explorer (WISE) surveys from SkyView Virtual Observatory along with SIMBAD Astronomical Database and Gaia Archive of ESA. The size of the cavity under study is 7.43 pc×2.90 pc, 2.99pc×0.99pc, and 1.26×0.53pc with an inclination angle of 71.12°, 75.73°, and 68.48° respectively in IRIS, AKARI and WISE data suggesting the cavity is neither a face-on nor an edge-on. The relative flux density of the region has been studied through pixel extraction of Far-Infrared (FIR) images and dust color temperature along with dust mass has been calculated. Using IRIS data, the temperature of the isolated region is found between a maximum value of 36.82±4.30 K to a minimum of 22.59±2.32 K with an offset of 14.23 K. The average temperature of the region is 28.22±0.18 K. Following similar procedures for AKARI data, the temperature is found between a maximum of 26.37±3.56 K and a minimum of 16.32±1.47 K with an offset of 10.06K. The average temperature is 19.25±0.15 K. Similarly, using WISE data, the temperature is found between a maximum of 353.72±18.54 K and a minimum of 307.24±4.69 K with an offset of 46.48K. The average temperature is 316.62±0.81 K. High value of offset temperature suggests that the cavity might be evolving with disruptions from background radiative sources. Approximately Gaussian distribution of the temperature in all the surveys implies that the region might be tending towards local thermodynamic equilibrium. The total mass of the structure is estimated to be around 0.03±1.5×10−6M⊙, 1.6×10−3±1.4×10−5 M⊙, and 10−7±3×10−10 M⊙ using IRIS, AKARI, and WISE data respectively. The Jean’s mass of the structure is calculated assuming the structure is a non-degenerate gas, which is estimated around 2001.93M⊙, 178.44M⊙, and 121.47M⊙ respectively using IRIS, AKARI, and WISE data. Since Jean’s mass is much greater than the mass of the structure there seems no possibility of star formation within the region of interest. The color maps illustrate identical distribution for all wavelengths, however, no significant relation is observed between dust color temperature and dust mass.

Preparation, structure and properties of bamboo charcoal and its modified polyester fiber

Bamboo charcoals (BC) are prepared at different pyrolysis temperature (350, 450, 500, 600 and 700 °C) with 2 h residence time and at 15 °C/min heating rates. The biochar yields, thermal stability, molecular groups, surface topography, and pore structure were characterized to explore the effect of pyrolysis temperatures on the characteristics of BCs. The results show that the biochar yield has decreased while the thermal stability of BC has increased with the increasing pyrolysis temperature. With the increase of temperature, C-C and C-H bond of aromatic compounds are enhanced to form stable components. The highest specific surface area and pore volume is obtained in the BC prepared at 500 °C with the value of 40.754 m2/g and 0.04738 cm3/g. The optimum temperature for achieving sufficient structure and good properties in fiber is 500 °C. And then, PET/BC fibers are prepared via melt-blend spinning. The mechanical capacity, thermal behavior, moisture absorption and far infrared emission properties of PET/BC fibers with different BC content have been investigated. It is found that the fiber has most desired comprehensive property with the BC content was 2%. At this composition, the tensile strength of the PET is 3.01 cN/dtex, the moisture absorption increases by 50%. The far-infrared imaging of the fabric shows highly temperatures raise of 44.8 °C within 30 s.

Capitalizing on RGB-FIR Hybrid Imaging for Road Detection

Traditionally, road detection approaches mostly capitalize on RGB images, 3D LiDAR point cloud or their fusion. However, RGB camera is sensitive to light conditions, while LiDAR point cloud is sparse compared with dense image pixels. In this work, a new hybrid image dataset is provided for the task of road detection based on cameras. In this dataset, the hybrid images are acquired by an optically aligned hybrid imaging device, consisting of a far-infrared (FIR) imager and an RGB camera to output pixel-wise registration of thermal and RGB frames. Then we investigate on three methods based on fully convolutional neural network (F-CNN) to demonstrate the advantages by fusing RGB-FIR images in road detection. First, a middle-fusion based model is built, where the output feature maps of encoder branches from RGB and FIR images are directly concatenated into a single-fusion branch as the decoder. Next, the originally discarded layers after fusion operation for both RGB and FIR branches are recovered as the mimic branches to imitate the distributions of the fusion outputs, which constitutes an extended cross model (ECM). Moreover, the outputs of mimic branches at different scales are also used to imitate the corresponding outputs in the fusion branch, called a hierarchical cross model (HCM). The experimental results demonstrate the effectiveness and efficiency of our fusion strategies.

Alignment and fusion of visible and infrared images based on gradient-domain processing

An image fusion of different modal images, such as visible and far-infrared images, is an important image processing technique because different modal images can compensate for each other. Many existing image fusion algorithms assume that different modal images are perfectly aligned. However, that assumption is not satisfied in many practical situations. In this paper, we propose an image alignment and fusion algorithm with gradient-domain processing. First, we extract the gradient maps from both modality images. Then, assuming disparities between the two gradient maps, candidate gradient maps for the target fused image are generated by selecting the gradient having larger power from different modality images pixel-by-pixel. A key observation is as follows. If the assumed disparity is wrong, the fused image includes ghost edges. If the assumed disparity is correct, the single edge is preserved without the ghost edge in the fused image. Therefore, we evaluate the gradient power in the region-of-interest of the fused image with different diparities. Then, we can align images based on the disparity associated with the minimum gradient power. Finally, we apply gradient-based image fusion with the aligned image pairs. We experimentally validate that the proposed approach can effectively align and fuse the visible and far-infrared images.

Research on the stitching method of massive aerial far-infrared images with an inconsistent overlap rate

Research on the stitching method of massive aerial far-infrared images with an inconsistent overlap rate

Research on Lightweight Infrared Pedestrian Detection Model Algorithm for Embedded Platform

Due to the limitation of energy consumption and power consumption, the embedded platform cannot meet the real-time requirements of the far-infrared image pedestrian detection algorithm. To solve this problem, this paper proposes a new real-time infrared pedestrian detection algorithm (RepVGG-YOLOv4, Rep-YOLO), which uses RepVGG to reconstruct the YOLOv4 backbone network, reduces the amount of model parameters and calculations, and improves the speed of target detection; using space spatial pyramid pooling (SPP) obtains different receptive field information to improve the accuracy of model detection; using the channel pruning compression method reduces redundant parameters, model size, and computational complexity. The experimental results show that compared with the YOLOv4 target detection algorithm, the Rep-YOLO algorithm reduces the model volume by 90%, the floating-point calculation is reduced by 93.4%, the reasoning speed is increased by 4 times, and the model detection accuracy after compression reaches 93.25%.

LIALFP: Multi-band images synchronous fusion model based on latent information association and local feature preserving

The fusion of multi-band images, i.e., far-infrared image (FIRI), near-infrared image (NIRI) and visible image (VISI), is mainly confronted with three-sided challenges: One is whether image fusion process could be performed synchronously. Most of existing algorithms are aimed at two fusion targets, which makes them have to adopt sequential way to merge multiple images. Unfortunately, this may make their results vulnerable to ambiguity or artifact. The second is the ground truth of fused results could not be obtained at all in some image fusion fields (e.g., multi-band images). That leads immediately to the failure of supervised methods to give full play to their advantages. Third, the latent projection between the result and source images is often not directly considered. Notably, the relation involves not only the fusion result with all inputs, but also with each original. In order to solve aforementioned problems, this paper establishes a unsupervised representation learning model for synchronous multi-band images fusion. First, significant pixel fusion features are extracted to ensure that the primary information can be integrated. Secondly, the potential relationship between the result and the whole originals is assumed to be the linear mapping, reducing the unpredictability of these fusion results. In addition, these transformation matrices have been given the function of feature selection, which could choose discriminant features and project them into the fusion space. Then, the locally significant features of each source are captured by designed graph Laplacian matrix. Finally, experiments show the rationality and superiority of our algorithm through comparison with a variety of recent advanced algorithms from subjective judgment and objective indicators.

V488 Per Revisited: No Strong Mid-infrared Emission Features and No Evidence for Stellar/substellar Companions

Abstract We present characterization of the planetary system architecture for V488 Per, the dustiest main-sequence star known with a fractional infrared luminosity of ≈16%. Far-infrared imaging photometry confirms the existence of an outer planetary system dust population with a blackbody-fit temperature of ≈130 K. Mid-infrared spectroscopy probing the previously identified ≈800 K inner planetary system dust population does not detect any obvious solid-state emission features, suggesting either large grain sizes that mute such emission and/or grain compositions dominated by species like amorphous carbon and metallic iron, which do not produce such features. In the latter case, the presence of significant quantities of iron-rich material could be indicative of the active formation of a Mercury-like planet around V488 Per. In any event, the absence of solid-state emission features is very unusual among main-sequence stars with copious amounts of warm orbiting dust particles; we know of no other such star whose mid-infrared spectrum lacks such features. Combined radial velocity monitoring and adaptive optics imaging find no evidence for stellar/substellar companions within several hundred astronomical units of V488 Per.

Study of Dust Cavity around the White Dwarf WD 0352-049 in Infrared Astronomical Satellite Map

In this work, we have studied the far-infrared images of the dust cavity around the White Dwarf WD 0352-049 available in Infrared Astronomical Satellite Map from Sky View Observatory. The size of the cavity is 24.48 pc × 8.10 pc. We have studied the relative infrared flux density and calculated the dust color temperature and dust mass. The temperature of the whole cavity structure lies between a maximum value 24.09 ± 0.50 K to a minimum 21.87 ± 0.61K with fluctuation of 2.22 K and an average value of 23.09 ± 1.11 K. The small fluctuation of dust color temperature suggests that the dust in cavity structure is evolving independently and less disturbed from background radiation sources. The color map shows the identical distribution of flux at 60 μm and 100 μm and the inverse distribution of dust color temperature and dust mass. There is a Gaussian-like distribution of relative flux density, dust color temperature and dust mass. The Gaussian distribution of temperature suggests that the dusts in cavity are in local thermodynamic equilibrium. The study of relative flux density and dust color temperature along the major and minor axis shows there is a sinusoidal fluctuation of flux and temperature, which might be due to the wind generated by White Dwarf located nearby the center of the cavity structure. The total dust mass of the dust is found to be 0.07 Mʘ and that of gas is 13.66 Mʘ. The Jeans mass of the structure is less than the total mass of gas in the structure, suggesting the possibility of star formation activity by gravitational collapse in the future. Also, the study of inclination angle suggests that the three-dimensional shape of the structure is uniform and regularly shaped.

The Galaxy Activity, Torus, and Outflow Survey (GATOS)

We compare high angular resolution mid-infrared (mid-IR) and Atacama Large Millimeter/submillimeter Array (ALMA) far-infrared (far-IR) images of twelve nearby (median 21 Mpc) Seyfert galaxies selected from the Galaxy Activity, Torus, and Outflow Survey (GATOS). The mid-IR unresolved emission contributes more than 60% of the nuclear (diameters of 1.5″ ∼ 150 pc) emission in most galaxies. By contrast, the ALMA 870 μm continuum emission is mostlyresolved with a median diameter of 42 pc and typically along the equatorial direction of the torus (Paper I). The Eddington ratios and nuclear hydrogen column densities (NH) of half the sample are favorable to launching polar and/or equatorial dusty winds, according to numerical simulations. Six of these show mid-IR extended emission approximately in the polar direction as traced by the narrow line region and perpendicular to the ALMA emission. In a few galaxies, the nuclear NH might be too high to uplift large quantities of dusty material along the polar direction. Five galaxies have low NH and/or Eddington ratios and thus polar dusty winds are not likely. We generated new radiative transfer CAT3D-WIND disk+wind models and model images at 8, 12, and 700 μm. We tailored these models to the properties of the GATOS Seyferts in this work. At low wind-to-disk cloud ratios, the far-IR model images have disk- and ring-like morphologies. The characteristic “X”-shape associated with dusty winds is seen better in the far-IR at intermediate-high inclinations for the extended-wind configurations. In most of the explored models, the mid-IR emission mainly comes from the inner part of the disk and cone. Extended biconical and one-sided polar mid-IR emission is seen in extended-wind configurations and high wind-to-disk cloud ratios. When convolved to the typical angular resolution of our observations, the CAT3D-WIND model images reproduce qualitative aspects of the observed mid- and far-IR morphologies. However, low to intermediate values of the wind-to-disk ratio are required to account for the observed large fractions of unresolved mid-IR emission in our sample. This work and Paper I provide observational support for the torus+wind scenario. The wind component is more relevant at high Eddington ratios and/or active galactic nucleus luminosities, and polar dust emission is predicted at nuclear column densities of up to ∼1024 cm−2. The torus or disk component, on the other hand, prevails at low luminosities and/or Eddington ratios.

3D Pedestrian Detection in Farmland by Monocular RGB Image and Far-Infrared Sensing

The automated driving of agricultural machinery is of great significance for the agricultural production efficiency, yet is still challenging due to the significantly varied environmental conditions through day and night. To address operation safety for pedestrians in farmland, this paper proposes a 3D person sensing approach based on monocular RGB and Far-Infrared (FIR) images. Since public available datasets for agricultural 3D pedestrian detection are scarce, a new dataset is proposed, named as “FieldSafePedestrian”, which includes field images in both day and night. The implemented data augmentations of night images and semi-automatic labeling approach are also elaborated to facilitate the 3D annotation of pedestrians. To fuse heterogeneous images of sensors with non-parallel optical axis, the Dual-Input Depth-Guided Dynamic-Depthwise-Dilated Fusion network (D5F) is proposed, which assists the pixel alignment between FIR and RGB images with estimated depth information and deploys a dynamic filtering to guide the heterogeneous information fusion. Experiments on field images in both daytime and nighttime demonstrate that compared with the state-of-the-arts, the dynamic aligned image fusion achieves an accuracy gain of 3.9% and 4.5% in terms of center distance and BEV-IOU, respectively, without affecting the run-time efficiency.

The peculiar nebula Simeis 57

Simeis 57 (HS 191) is an optically bright nebula in the Cygnus X region with a peculiar appearance that suggests an outflow from a rotating source. Newly obtained observations and archival data reveal Simeis 57 as a low-density (ne ~ 100 cm−3) nebula with an east-to-west excitation gradient. The extinction of the nebula is AV ≤ 2 mag. The nebula is recognizable but not prominent in mid- and far-infrared images. In its direction, half a dozen small CO clouds have been identified at VLSR = +5 km s−1. One of these coincides with both the optical nebula and a second CO cloud at the nebular velocity VLSR ≈−10 km−1. No luminous stars are embedded in these molecular clouds, nor are any obscured by them and no sufficiently luminous stars are found in the immediate vicinity of the nebula. Instead, all available data points to the evolved star HD 193793 = WR 140 (an O4-5 supergiant and WC7 Wolf–Rayet binary) as the source of excitation, notwithstanding its large separation of 50′, about 25 pc at the stellar distance of 1.7 kpc. Simeis 57 appears to be a part of a larger structure surrounding the HI void centered on HD 193793.

Pose estimation at night in infrared images using a lightweight multi-stage attention network

Human Keypoints Detection is a relatively basic task in computer vision; it is the pre-task of human action recognition, behavior analysis and human–computer interaction. Since most abnormal actions occur at night, how to effectively extract skeleton sequence data in a low-light or completely dark environment poses a huge challenge for its identification. This paper proposes to use far infrared images to detection key points of the human body, which can solve the problem of human pose estimation under challenging weather conditions such as total darkness, smoke, inclement weather and glare. However, far-infrared images have some shortcomings, such as low resolution, noise and thermal characteristics; the skeleton data need to be provided in real time for the next stage of task. Based on the above reasons, this paper proposes a lightweight multi-stage attention network (LMANet) to detect the key points of human at night. This new network structure adds context information through the large receptive field, which helps to assist the detection of neighboring key points through this information, but for the sake of lightweight consideration, this article only extends the network to two stages. In addition, this article uses the attention module to effectively select channels with a large amount of information and highlight the features of key points, while eliminating background interference. In order to detect key points of the human in various complex environments, we use techniques such as difficult sample mining which improves the accuracy of key points with low confidence. Our network has been verified on two visible light datasets, fully demonstrating excellent performance. This paper successfully introduces far-infrared images into the field of pose estimation, because there is no public dataset for far-infrared pose estimation. In this paper, 700 images are selected for annotation from multiple public far-infrared object detection, segmentation and action recognition datasets; our algorithm is verified on this dataset; the effect is very good. After the paper is published, we will publish our key points of the human body annotated documents.