Installing

After cloning this directory, run: sudo python setup.py install

Running

Usage ./VideoPipeline.py config.yaml video.file1 video.file2 video.file3 etc

Model Directory Requirement

Models must be placed in the CWD for VideoPipeline with the following structure: Eg. for washingDishing version 42 and 73, we expect to see the following files:

structSVM-data/datasets/washingDishes structSVM-data/datasets/washingDishes/models structSVM-data/datasets/washingDishes/models/classes.txt.codebook.0 structSVM-data/datasets/washingDishes/models/modelGraph.svg structSVM-data/datasets/washingDishes/models/classes.txt.codebook.5 structSVM-data/datasets/washingDishes/models/train.txt structSVM-data/datasets/washingDishes/models/svm_params.txt structSVM-data/datasets/washingDishes/models/classes.txt structSVM-data/datasets/washingDishes/models/classes.txt.codebook.10 structSVM-data/datasets/washingDishes/models/model.txt structSVM-data/datasets/washingDishes/models/modelGraph.m structSVM-data/datasets/washingDishes/models/test.results.txt structSVM-data/datasets/washingDishes/models/test.txt structSVM-data/datasets/washingDishes.42 structSVM-data/datasets/washingDishes.42/models structSVM-data/datasets/washingDishes.42/models/classes.txt.codebook.0 structSVM-data/datasets/washingDishes.42/models/modelGraph.svg structSVM-data/datasets/washingDishes.42/models/classes.txt.codebook.5 structSVM-data/datasets/washingDishes.42/models/train.txt structSVM-data/datasets/washingDishes.42/models/svm_params.txt structSVM-data/datasets/washingDishes.42/models/classes.txt structSVM-data/datasets/washingDishes.42/models/classes.txt.codebook.10 structSVM-data/datasets/washingDishes.42/models/model.txt structSVM-data/datasets/washingDishes.42/models/modelGraph.m structSVM-data/datasets/washingDishes.42/models/test.results.txt structSVM-data/datasets/washingDishes.42/models/test.txt structSVM-data/datasets/washingDishes.73 structSVM-data/datasets/washingDishes.73/models structSVM-data/datasets/washingDishes.73/models/classes.txt.codebook.0 structSVM-data/datasets/washingDishes.73/models/modelGraph.svg structSVM-data/datasets/washingDishes.73/models/classes.txt.codebook.5 structSVM-data/datasets/washingDishes.73/models/train.txt structSVM-data/datasets/washingDishes.73/models/svm_params.txt structSVM-data/datasets/washingDishes.73/models/classes.txt structSVM-data/datasets/washingDishes.73/models/classes.txt.codebook.10 structSVM-data/datasets/washingDishes.73/models/model.txt structSVM-data/datasets/washingDishes.73/models/modelGraph.m structSVM-data/datasets/washingDishes.73/models/test.results.txt structSVM-data/datasets/washingDishes.73/models/test.txt

Config

config.yaml contains the plugins and the models that will be executed Each ModelDetection config is listed as such

ModelDetection0:
    defaultOrder     : 6
    threshold        : 0.6
    stronger         : 0.2
    weaker           : -0.2
    modelName        : "skiingBoarding"
    modelVersion     : 59

ModelDetection[0-1000] are checked

Plugins

FrameExtraction, BlankDetection, SelectingSingleFrame, RemoveMultiModels, ModelDetection are run

Results

Results are stored in /results/.json Eg. BlankDetection.json, ModelDetection:washingDishes:73.json, ModelDetection:washingDishes:42.json, etc

Result Format for each Plugin File: "FrameId": { "Score": "Value", "State": "Value" }, "FrameId1": { ... }, ....

Frames generated

Generated frames are in ppm format and are stored at /frames//original.ppm When features are extracted they are placed alongside the frame img with the .feat.bin extension

Libraries Required

libzsvm.so from ImageClassifier.cpp

zigvu / khajuri

readme