Module vipy.data.activitynet

Classes

class ActivityNet (datadir='/Users/jebyrne/.vipy/activitynet', redownload=False)

http://activity-net.org

Activitynet, provide a datadir='/path/to/store/activitynet'

Expand source code Browse git

class ActivityNet():
    """http://activity-net.org"""
    def __init__(self, datadir=vipy.util.tocache('activitynet'), redownload=False):
        """Activitynet, provide a datadir='/path/to/store/activitynet' """
        self._url = URL
        self.datadir = remkdir(datadir)
        if redownload or not self._isdownloaded():
            self.download()
        
    def __repr__(self):
        return str('<vipy.data.activitynet: "%s">' % self.datadir)

    def download(self):
        vipy.downloader.download(URL, os.path.join(self.datadir, filetail(URL)))
        return self

    def _isdownloaded(self):
        return os.path.exists(os.path.join(self.datadir, 'activity_net.v1-3.min.json'))
    
    def _dataset(self, subset):
        assert self._isdownloaded(), "Dataset not downloaded.  download() first or manually download '%s' into '%s'" % (self._url, self.datadir)        
        jsonfile = os.path.join(self.datadir, filetail(URL))
        json = readjson(jsonfile)

        return [(v['url'],
                 os.path.join(self.datadir, youtubeid),
                 a['label'],
                 float(a['segment'][0]),
                 float(a['segment'][1]))
                for (youtubeid, v) in json['database'].items()
                for a in v['annotations']
                if v['subset'] == subset]

    def trainset(self):
        loader = lambda x: VideoCategory(url=x[0], filename=x[1], category=x[2], startsec=x[3], endsec=x[4], framerate=None)
        return vipy.dataset.Dataset(self._dataset('training'), id='activitynet:train', loader=loader)

    def testset(self):
        """ActivityNet test set does not include any annotations"""
        assert self._isdownloaded(), "Dataset not downloaded.  download() first or manually download '%s' into '%s'" % (self._url, self.datadir)        
        json = readjson(os.path.join(self.datadir, filetail(URL)))
        loader = lambda x: Video(url=x[0], filename=x[1], framerate=None)
        return vipy.dataset.Dataset([(v['url'], os.path.join(self.datadir, youtubeid)) for (youtubeid, v) in json['database'].items() if v['subset'] == 'testing'], id='activitynet:test', loader=loader)

    def valset(self):
        loader = lambda x: VideoCategory(url=x[0], filename=x[1], category=x[2], startsec=x[3], endsec=x[4], framerate=None)
        return vipy.dataset.Dataset(self._dataset('validation'), id='activitynet:val', loader=loader)
        
    def categories(self):
        return set([v.category() for v in self.trainset()])

    def analysis(self):
        C = self.categories()
        d_category_to_trainsize = {k:len(v) for (k,v) in groupbyasdict(self.trainset(), lambda x: x.category()).items()}

        top10 = sorted([(k,v) for (k,v) in d_category_to_trainsize.items()], key=lambda x: x[1])[-10:]
        print('top10 categories by number of instances in training set:')
        print(top10)

        bottom10 = sorted([(k,v) for (k,v) in d_category_to_trainsize.items()], key=lambda x: x[1])[0:10]
        print('bottom-10 categories by number of instances in training set:')
        print(bottom10)
        
        return d_category_to_trainsize

Methods

def analysis(self)

Expand source code Browse git

def analysis(self):
    C = self.categories()
    d_category_to_trainsize = {k:len(v) for (k,v) in groupbyasdict(self.trainset(), lambda x: x.category()).items()}

    top10 = sorted([(k,v) for (k,v) in d_category_to_trainsize.items()], key=lambda x: x[1])[-10:]
    print('top10 categories by number of instances in training set:')
    print(top10)

    bottom10 = sorted([(k,v) for (k,v) in d_category_to_trainsize.items()], key=lambda x: x[1])[0:10]
    print('bottom-10 categories by number of instances in training set:')
    print(bottom10)
    
    return d_category_to_trainsize

def categories(self)

Expand source code Browse git

def categories(self):
    return set([v.category() for v in self.trainset()])

def download(self)

Expand source code Browse git

def download(self):
    vipy.downloader.download(URL, os.path.join(self.datadir, filetail(URL)))
    return self

def testset(self)

ActivityNet test set does not include any annotations

Expand source code Browse git

def testset(self):
    """ActivityNet test set does not include any annotations"""
    assert self._isdownloaded(), "Dataset not downloaded.  download() first or manually download '%s' into '%s'" % (self._url, self.datadir)        
    json = readjson(os.path.join(self.datadir, filetail(URL)))
    loader = lambda x: Video(url=x[0], filename=x[1], framerate=None)
    return vipy.dataset.Dataset([(v['url'], os.path.join(self.datadir, youtubeid)) for (youtubeid, v) in json['database'].items() if v['subset'] == 'testing'], id='activitynet:test', loader=loader)

def trainset(self)

Expand source code Browse git

def trainset(self):
    loader = lambda x: VideoCategory(url=x[0], filename=x[1], category=x[2], startsec=x[3], endsec=x[4], framerate=None)
    return vipy.dataset.Dataset(self._dataset('training'), id='activitynet:train', loader=loader)

def valset(self)

Expand source code Browse git

def valset(self):
    loader = lambda x: VideoCategory(url=x[0], filename=x[1], category=x[2], startsec=x[3], endsec=x[4], framerate=None)
    return vipy.dataset.Dataset(self._dataset('validation'), id='activitynet:val', loader=loader)