def __init__(self, folder):
self.frames_list = list_paths(folder)
self.ind = 0
self.done = False
self.first = True
self.sequence_list = []
def detectFromFolder(folder): paths = list_paths(folder) images, locations, new_paths = detectFromPaths(paths) return images, locations, new_paths
def load_stadiums():
print("Loading stadiums.")
p = os.path.join(ROOT_DIR, 'metadata/data/places/stadiums')
pys = [e for e in list_paths(p) if e.endswith('.py')]
l = []
# What is this?
for py in pys:
if not py.startswith('_'):
#tail = py.replace('/home/chris/www/', '').replace('.py', '').replace('/', '.')
tail = py.replace(ROOT_DIR, '').replace('.py', '').replace('/', '.')
if tail.startswith('.'):
tail = tail[1:]
l.extend(import_path(tail).l)
regions = [e for e in os.listdir(p) if '.' not in e]
final = []
# Do this in normalize?
for e in l:
d = stadium_defaults.copy()
d.update(e)
final.append(d)
return final
async def index(request):
stages = {}
global imports
for (key, module) in imports.items():
setup = getattr(module, 'setup')
module_pys = get_import_files(setup['path'])
stages[key] = list_paths(module_pys)
return json({'success': True, 'data': stages})
async def stage_index(request, stage_name):
lessons = {}
global imports, module_pys
if stage_name in imports:
setup = getattr(imports[stage_name], 'setup')
lessons = list_paths(module_imports[stage_name])
return json({'success': True, 'data': lessons})
else:
return json({'fail': True, 'error': 'No such stage.'})
def test(self):
with torch.no_grad():
self.g_model.eval()
test_videos = list_paths(c.TEST_DIR)
psnr_all_videos = []
for vid in test_videos: # test videos seperately
vid_name = vid.split('/')[-1]
video_loader = Loader(folder=vid)
psnr_mean = self.process_video(video_loader)
print('{:15s} - PSNR: {:5.2f}'.format(vid_name, psnr_mean))
psnr_all_videos.append(psnr_mean)
psnr_videos_mean = np.mean(psnr_all_videos)
print('{:15s} - PSNR: {:5.2f}'.format('AVERAGE', psnr_videos_mean))
print('\nPredicted frames are saved at {}'.format(c.IMG_SAVE_DIR))