使用本地视频，用Openface软件进行面部特征分析的方法：
单个文件中提取：
assuming that only one person appears in that video file, execute the following command on the command line:
命令：FeatureExtraction.exe -f "C:\my videos\video.avi"
If there are multiple people in the video use:
命令：FaceLandmarkVidMulti.exe -f "C:\my videos\video.avi"
会创建一个过程目录在当前同样的工作目录中，会包含过程特征，更多细节
This will create a ‘processed’ directory in the same directory (working directory) as FeatureExtraction or FaceLandmarkVidMulti that will contain the processed features (more details).
多个文件中提取：
FeatureExtraction.exe -f "C:\my videos\video1.avi" -f "C:\my videos\video2.avi" -f "C:\my videos\video3.avi"
FaceLandmarkVidMulti.exe -f "C:\my videos\video1.avi" -f "C:\my videos\video2.avi" -f "C:\my videos\video3.avi"
如果只想提取头部姿势：
FeatureExtraction.exe -f "C:\my videos\video1.avi" -pose
如果一个文件中有多张图片想提取：
FeatureExtraction.exe -fdir "C:\my videos\sequence1" 适用于多张图片来源于一个video,且只含一张人脸
FaceLandmarkVidMulti.exe -fdir "C:\my videos\sequence1" 适用于图片之间没有什么关系，可以有多张人脸
从单张图片中提取：
FaceLandmarkImg.exe -f "C:\my images\img.jpg"
从多张图片中提取：
FaceLandmarkImg.exe -f "C:\my images\img1.jpg" -f "C:\my images\img2.jpg" -f "C:\my images\img3.jpg"
如果想使用某个目录下的所有图片：
FaceLandmarkImg.exe -fdir "C:\my images"
这个命令会使用所有这些格式的图片 (.jpg, jpeg, .bmp, .png files)
Another good way to see how the command line arguments work is through looking at ./matlab_runners/Demos/ which illustrates how each of the models is called.
网友评论