This project aims to implement video processing on Hadoop. We cutomize the FileInputFormat and RecordReader class in Hadoop. Please follow the following steps to run a demo job.
###Step 1 Clone the repository.
git clone [email protected]:xmpy/video-hadoop-template.git
###Step 2 Build this project.
mvn assembly:assembly
###Step 3 Run the jar.
hadoop jar target/metadata-0.0.1-SNAPSHOT.jar edu.bupt.videodatacenter.input.MRVideoReader /user/zhaoxm/vehicle_1min.avi /user/zhaoxm/metadata_output2
The last two argument are the input video path and output path.
See result. First column is the key, which is filename_framecount. Second column is the metadata of this frame image.
vehicle_1min.avi_1 BufferedImage@19ed13da: type = 5 ColorModel: #pixelBits = 24 numComponents = 3 color space = java.awt.color.ICC_ColorSpace@1bb25a82 transparency = 1 has alpha = false isAlphaPre = false ByteInterleavedRaster: width = 640 height = 360 #numDataElements 3 dataOff[0] = 2
vehicle_1min.avi_10 BufferedImage@19412332: type = 5 ColorModel: #pixelBits = 24 numComponents = 3 color space = java.awt.color.ICC_ColorSpace@1bb25a82 transparency = 1 has alpha = false isAlphaPre = false ByteInterleavedRaster: width = 640 height = 360 #numDataElements 3 dataOff[0] = 2