Office Meeting Person Recognition Dataset

In modern office environments, meeting efficiency depends on accurate person recognition. However, traditional methods are often limited by factors such as lighting changes and angle variations, leading to decreased recognition accuracy. Existing solutions lack coverage of various postures and lighting conditions in office scenarios, making it difficult to meet the demands of enterprise daily office automation systems. This dataset aims to provide high-precision person recognition data for the intelligent office meeting field, addressing the challenges of facial recognition in complex and dynamic environments. The dataset is collected through high-definition cameras in real meeting room environments, ensuring the inclusion of images of people with various postures, expressions, and accessories (such as glasses and hats). The data has undergone multiple rounds of professional annotation and consistency checks to ensure label accuracy is above 98%. The annotation team consists of 10 image processing experts with over 5 years of experience. Data preprocessing includes steps such as image enhancement and denoising to improve model robustness. The data is stored as well-organized JPG format files, facilitating quick retrieval and processing.