#ViPar Toolkit

Scrape, process, and filter Twitter's Vine video service to create unique video datasets on-the-fly.

The code in this respository should be used in the following order:

###1. VineScrape Collects annotated links to Vine videos from the Twitter Streaming API and generates linguistic meta-data. Requires a Twitter developer account.

###2. ImageNet Word list Generator Creates word lists of ImageNet hyponyms based on a user-specified category word. Need to generate list prior to running VineFilter.

###3. VineFilter Filters Vine video text descriptions by their linguistic meta-data (ie: parts-of-speech and grammar relationships) in order to return videos containing a high-probability of having the user-specified visual object/entity in the video content itself. The real power of the ViPar toolkit is demonstrated in this tool.

###4. VineFetch Downloads Vine videos from URLs stored in a user-specified JSON file.

###5. VineAnalyze Easily view and annotate Vine videos downloaded by VineFetch. Much faster than individually opening files.