from evaluate import mAP
import misc.h5_dataset as h5_dataset
opt = parse_args()
os.environ['CUDA_VISIBLE_DEVICES'] = str(opt.gpu)
if opt.h5:
trainset = h5_dataset.H5Dataset(opt, split=0)
valset = h5_dataset.H5Dataset(opt, split=1)
testset = h5_dataset.H5Dataset(opt, split=2)
opt.num_workers = 0
else:
if opt.dataset.find('iiit_hws') > -1:
trainset = datasets.SegmentedDataset(opt, 'train')
else:
trainset = datasets.Dataset(opt, 'train')
valset = datasets.Dataset(opt, 'val')
testset = datasets.Dataset(opt, 'test')
sampler = datasets.RandomSampler(trainset, opt.max_iters)
trainloader = DataLoader(trainset,
batch_size=1,
sampler=sampler,
num_workers=opt.num_workers)
valloader = DataLoader(valset, batch_size=1, shuffle=False, num_workers=0)
testloader = DataLoader(testset, batch_size=1, shuffle=False, num_workers=0)
torch.set_default_tensor_type('torch.FloatTensor')
torch.manual_seed(opt.seed)
torch.cuda.manual_seed(opt.seed)
torch.cuda.device(opt.gpu)
with open("botany_konz_eval/data/%s_results_%s.xml" % (dataset, mode),
'wb') as f:
newdoc.writexml(f, addindent=' ', newl='\n', encoding='utf-8')
#%%
opt = parse_args()
opt.augment = 0
os.environ['CUDA_VISIBLE_DEVICES'] = str(opt.gpu)
if opt.h5:
testset = h5_dataset.H5Dataset(opt, split=2)
valset = h5_dataset.H5Dataset(opt, split=1)
opt.num_workers = 0
else:
testset = datasets.Dataset(opt, 'test')
valset = datasets.Dataset(opt, 'val')
loader = dataloader.DataLoader(testset,
batch_size=1,
shuffle=False,
num_workers=0)
valloader = dataloader.DataLoader(valset,
batch_size=1,
shuffle=False,
num_workers=0)
torch.set_default_tensor_type('torch.FloatTensor')
torch.manual_seed(opt.seed)
torch.cuda.manual_seed(opt.seed)
torch.cuda.device(opt.gpu)