PHOENIX14T (RWTH-PHOENIX-Weather-2014T)

签名由放置在手语翻译前的固定彩色摄像机记录下来。口译员在带有颜色过渡的人造灰色背景前穿着深色衣服。所有录制的视频均为每秒 25 帧，帧大小为 210 x 260 像素。每个框架仅显示解释器框。

The sign language performances are recorded by a fixed color camera placed in front of the sign language interpreter. The interpreter wears dark-colored clothing and stands in front of a synthetic gray background with color transitions. All recorded videos have a frame rate of 25 frames per second, with a frame size of 210 × 260 pixels. Each frame only displays the interpreter's bounding box.

搜集汇总

背景与挑战

背景概述

PHOENIX14T是由亚琛工业大学2018年发布的手语翻译数据集，包含固定摄像机拍摄的25fps、210x260像素的手语翻译视频，每个视频帧仅显示解释器框。该数据集适用于手语识别和翻译研究。