法庭研究中视频篡改数据集的计算机辅助标注
Computer-Aided Annotation for Video Tampering Dataset of Forensic Research
摘要：
视频篡改数据集的标注是一项耗费大量人力和财力的枯燥工作。目前，还没有公开的文献能够提高伪造视频的标注效率。提出了一种视频篡改数据集的计算机辅助标注方法。该方法可用于对伪造视频序列的帧进行标注。通过比较原始视频帧和伪造视频帧，可以定位伪造视频帧中伪造区域的位置和轨迹。然后，根据伪造区域的运动轨迹在时域上选取几个关键点，并在关键点处用鼠标标记出伪造帧的伪造区域。最后，利用基于时间域关键位置坐标的线性预测算法，生成其他视频帧中伪造区域的标注信息，无需人工标注。如果计算机辅助算法生成的限界框偏离了伪造区域的实际位置，我们可以在预览期间使用鼠标改变限界框的位置。该方法将人工标注与计算机辅助标注相结合。该方法解决了计算机辅助标注的不准确性和人工标注效率低的问题，满足了视频被动取证研究中对大量伪造视频进行标注的需要。
英文摘要：
The annotation of video tampering dataset is a boring task that takes a lot of manpower and financial resources. At present, there is no published literature which is capable to improve the annotation efficiency of forged videos. We presented a computer-aided annotation method for video tampering dataset in this paper. This annotation method can be utilized to label the frames of forged video sequences. By means of comparing the original video frames with the forged video frames, we can locate the position and the trajectory of the forged areas of the forged video frames. Then, we select several key points on the temporal domain according to the trajectory of the forged areas, and mark the forged area of the forged frames in the key point with a mouse. Finally, we use the linear prediction algorithm based on the coordinates of the key positions in the temporal domain to generate the annotation information of forged areas in other video frames which without manually labeled. If the bounding box generated by the computer-aided algorithm deviates from the actual location of the forged area, we can use the mouse to change the position of the bounding box during the preview period. This method combines the manual annotation with computer-aided annotation. It solves the problems of the inaccuracy of annotation by computer-aided as well as the low efficiency of annotation manually, and meet the needs of annotation for an enormous amount of forged videos in the research of video passive forensics.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
