High-Quality Multi-Modal Face Dataset to Empower Face Recognition Models

Professional-Grade Dataset with Multi-Angle Face Images, Videos and Audio

• Leveraging years of expertise in image data processing, Surfing Tech provides high-quality datasets tailored for training and testing various facial recognition tasks to AI practitioners. The datastes contains 300 subjects, with each reciting 300 sentences. The recording setup includes 1 lapel mic for audio, 1 smartphone for audio, and another smartphone for video, which are manually synced.

• The audio captured by smartphone has a sampling rate of 44.1KHz at 16bit depth. The video is recorded at 1440¡Á1080 resolution at 60FPS frame rate. The audio from lapel mic is also sampled at 44.1KHz with 16bit depth. The three modalities are aligned manually into synchronous audio-visual data.

• Capturing speech and facial expressions across a range of emotions, this dataset enables building robust multi-angle face recognition models with improved adaptability. 

• We welcome AI developers to use our face dataset solutions for expedited iterations and upgrades of facial recognition capabilities.