class SegmentationInspector:
"""
For image binnary segmentation tasks, create a HTML report with best segmented
samples, worst segmented samples, and samples closest to the boundary
decision.
Args:
nb_show (int): how many samples to show
classes (list of str): classes names
post_each_batch (bool) whether to generate a new report on each batch
or only on epoch end.
"""
def __init__(self, nb_show, classes, post_each_batch=True):
self.vis = SIVis(nb_show, classes, 0.5)
self.post_each_batch = post_each_batch
def on_epoch_start(self, state):
if 'report' in state['metrics']:
del state['metrics']['report']
self.vis.reset()
@torch.no_grad()
def on_batch_end(self, state):
pred, y, x = state['pred'], state['batch'][1], state['batch'][0]
self.vis.analyze(x, pred, y)
if self.post_each_batch and state.get('visdom_will_log', False):
state['metrics']['report'] = self.vis.show()
def on_epoch_end(self, state):
state['metrics']['report'] = self.vis.show()
def __init__(self, nb_show, classes, post_each_batch=True):
self.vis = SIVis(nb_show, classes, 0.5)
self.post_each_batch = post_each_batch