A novel database of Teenager's spontaneous facial expressions

This dataset presents a comprehensive collection of spontaneous facial expressions obtained from 14 teenagers, encompassing six fundamental emotions: Anger, Disgust, Fear, Happy, Sad, and Surprise. With a total of 314 videos, the dataset comprises approximately 59,000 emotional frames, including both emotional expressions and neutral states. Recorded using an HD Webcam with 1080P resolution and a frame rate of 30 fps, the dataset captures authentic and unposed facial expressions, distinguishing it from existing databases. Its diverse subjects from various backgrounds enable cross-cultural and ethnic studies of facial expressions and emotions. Addressing the scarcity of similar datasets focused on teenagers, the dataset underwent thorough validation by seven validators, achieving a 64% average accuracy in emotion recognition. The dataset's potential applications range from enhancing facial expression recognition systems to investigating the impact of facial and body movements, while also providing a valuable resource for benchmarking and comparison within the vision community.本数据集收录了来自14名青少年的自然自发面部表情全面样本，涵盖愤怒（Anger）、厌恶（Disgust）、恐惧（Fear）、快乐（Happy）、悲伤（Sad）与惊讶（Surprise）六种基础情绪。数据集共包含314段视频，总计约59000帧表情画面，涵盖情绪表达与中性面部状态两类样本。该数据集采用1080P分辨率、30fps帧率的高清网络摄像头录制，采集的均为真实自然的非摆拍面部表情，这使其区别于现有公开数据库。由于受试者背景多元，本数据集支持针对面部表情与情绪开展跨文化、跨种族的相关研究。针对当前青少年主题同类数据集稀缺的现状，本数据集经过7名校验人员的严格审核，在情绪识别任务中平均准确率可达64%。该数据集的应用场景广泛，既可用于优化面部表情识别系统、探究面部与肢体动作的影响机制，也能为计算机视觉领域的基准测试与算法对比提供宝贵的研究资源。