Dataset of thermal facial videos of people breathing while wearing masks in different methods

Dataset of thermal facial videos of people breathing while wearing masks in different methods - Thermal video of the face for 4 seconds of a single breath while wearing a mask. - Five repetitions of video with the same conditions. - 5 different mask types used. (Labeled in the filename with a number as shown below.) Surgical Mask: Mask for surgical use KF-AD (Anti-Droplet): Mask for blocking droplets (approximately 5 μm in size) KF80: Blocks at least 80% of particles with an average size of 0.6 μm (similar rating to European FFP1) KF94: Blocks at least 94% of particles with an average size of 0.4 μm (similar to European FFP2 and U.S. N95) KF99: Blocks at least 99% of particles with an average size of 0.4 μm - 4 different methods of wearing a mask used. (Labeled in the filename with a number as shown below.) Wearing a mask on the chin to expose the nose and mouth Lowering the mask under the nose and covering only the mouth Covering the nose without the nose wire being tightly attached, thereby letting air leak from the side of the mask Wearing the mask properly and securely - File name: 'Participant number_Mask type number_Mask wearing method number_Repetition count' - The videos are in mp4 format with 1440×1080 pixels and frame rate of 8.7 Hz. - Organized into folders: train (3200 videos), test (1000 videos), validation (800 videos).

不同口罩佩戴方式下佩戴口罩人群呼吸过程中的面部热成像视频数据集。该数据集包含单一次呼吸过程中采集的4秒面部热成像视频，相同实验条件下的视频共重复采集5次。共使用5种不同类型的口罩（文件名以数字标注，规则如下）：
1. 外科口罩（Surgical Mask）：外科手术专用口罩
2. KF-AD（Anti-Droplet）：防飞沫口罩，可阻挡粒径约5μm的飞沫
3. KF80：可阻挡至少80%平均粒径为0.6μm的颗粒物，防护等级等效于欧洲FFP1标准
4. KF94：可阻挡至少94%平均粒径为0.4μm的颗粒物，防护等级等效于欧洲FFP2及美国N95标准
5. KF99：可阻挡至少99%平均粒径为0.4μm的颗粒物

共采用4种不同的口罩佩戴方式（文件名以数字标注，规则如下）：
1. 将口罩佩戴于下巴处，暴露鼻腔与口腔
2. 将口罩下拉至鼻尖下方，仅遮盖口腔
3. 未压紧鼻梁条佩戴口罩，导致空气从口罩侧边泄漏
4. 规范且紧密地佩戴口罩

文件命名格式为："Participant number_Mask type number_Mask wearing method number_Repetition count"。所有视频均采用mp4格式，分辨率为1440×1080像素，帧率为8.7Hz。数据集按文件夹划分为训练集（train，共3200个视频）、测试集（test，共1000个视频）与验证集（validation，共800个视频）。