In the future, this file name will likely be timestamped by the iLet app.
Run
python ensemble.py "test.mov" "test.wav" "test_registry.txt"
Where the arguments are the name of the video you just made, a name for the corresponding audio file, and a name for a registry that maps video segment file to sound classification.
You may need to change low,high (the number of segments to split the video into) to get the proper classification analysis. Hopefully, we will think of a way to code this in!
3. Analyze the video pictures! Classify what's in each video segment and store it for the iLet app to use later.
Run
python sample.py --textfile "test_registry.txt" "test_tags.txt"
Where the arguments are the registry file from above and a name for the image classifications within each video segment.
To analyze video snippets, we processed screenshots of the split files with PyTorch. The image captioning tutorial is made by @yunjey. Following the instructions detailed in the Usage, we can then test the model by calling:
$ python sample.py --image='png/example.png'