SpeakingFaces

speakFaces 是一个公开可用的大型数据集，旨在支持在结合了热、视觉和音频数据流的环境中进行多模式机器学习研究；示例包括人机交互 (HCI)、生物特征认证、识别系统、域转移和语音识别。 speakFaces 由完全对齐的高分辨率热和视觉光谱图像流组成，这些图像流与每个主题的音频记录同步，每个主题说出大约 100 个命令短语。

speakFaces is a publicly available large-scale dataset designed to support multimodal machine learning research in scenarios integrating thermal, visual and audio data streams. Examples include human-computer interaction (HCI), biometric authentication, recognition systems, domain shift, and speech recognition. speakFaces consists of fully aligned high-resolution thermal and visual spectral image streams, which are synchronized with audio recordings for each subject, and each subject utters approximately 100 command phrases.