def voc_ap(model, cfg):
imdb_name = 'voc_2007_test'
output_dir = 'models/testing/' + cfg['exp_name']
if not os.path.exists(output_dir):
os.makedirs(output_dir)
imdb = VOCDataset(imdb_name, '../data', cfg['batch_size'],
yolo_utils.preprocess_test, processes=4, shuffle=False, dst_size=cfg['inp_size'])
net = Darknet19(cfg)
net_utils.load_net(model, net)
net.cuda()
net.eval()
mAP = eval_net(net, cfg, imdb, max_per_image=300, thresh=0.001, output_dir=output_dir, vis=False)
imdb.close()
return mAP
def test_voc_ap(model):
print(model)
imdb = VOCDataset(imdb_name,
cfg.DATA_DIR,
cfg.batch_size,
yolo_utils.preprocess_test,
processes=4,
shuffle=False,
dst_size=cfg.inp_size)
net = Darknet19()
net_utils.load_net(model, net)
net.cuda()
net.eval()
mAP = test_net(net, imdb, max_per_image, thresh, vis)
imdb.close()
return mAP
import utils.yolo as yolo_utils
import utils.network as net_utils
from utils.timer import Timer
import cfgs.config as cfg
from random import randint
try:
from tensorboardX import SummaryWriter
except ImportError:
SummaryWriter = None
# data loader
imdb = VOCDataset(cfg.imdb_train,
cfg.DATA_DIR,
cfg.train_batch_size,
yolo_utils.preprocess_train,
processes=2,
shuffle=True,
dst_size=cfg.multi_scale_inp_size)
# dst_size=cfg.inp_size)
print('load data succ...')
net = Darknet19()
# net_utils.load_net(cfg.trained_model, net)
pretrained_model = os.path.join(cfg.train_output_dir,
'darknet19_voc07trainval_exp3_17.h5')
#pretrained_model = cfg.trained_model
net_utils.load_net(pretrained_model, net)
#net.load_from_npz(cfg.pretrained_model, num_conv=18)
net.cuda()
net.train()
cfg,
thr=0.1)
if im2show.shape[0] > 1100:
im2show = cv2.resize(im2show,
(int(1000. * float(im2show.shape[1]) / im2show.shape[0]), 1000)) # noqa
cv2.imshow('test', im2show)
cv2.waitKey(0)
with open(det_file, 'wb') as f:
pickle.dump(all_boxes, f, pickle.HIGHEST_PROTOCOL)
print('Evaluating detections')
imdb.evaluate_detections(all_boxes, output_dir)
if __name__ == '__main__':
# data loader
imdb = VOCDataset(imdb_name, cfg.DATA_DIR, cfg.batch_size,
yolo_utils.preprocess_test,
processes=1, shuffle=False, dst_size=cfg.multi_scale_inp_size)
net = Darknet19()
net_utils.load_net(trained_model, net)
net.cuda()
net.eval()
test_net(net, imdb, max_per_image, thresh, vis=False) #yuanlai shi false
imdb.close()
box_regression = box_regression.reshape(N, -1, 4)
box_loss = det_utils.smooth_l1_loss(
box_regression[sampled_pos_inds_subset, labels_pos],
regression_targets[sampled_pos_inds_subset],
beta=1 / 9,
size_average=False,
)
box_loss = box_loss / labels.numel()
return classification_loss, box_loss
# dataset_train = VOCDataset(root='/home/neuroplex/data/VOCdevkit/VOC2007')
dataset_train = VOCDataset(
root=
'/Users/pranoyr/code/Pytorch/faster-rcnn.pytorch/data/VOCdevkit2007/VOC2007'
)
dataloader = DataLoader(dataset_train,
num_workers=0,
collate_fn=collater,
batch_size=1)
# dataset_train = VRDDataset('/Users/pranoyr/code/Pytorch/faster-rcnn.pytorch/data/VRD', 'train')
# dataloader = DataLoader(
# dataset_train, num_workers=0, collate_fn=collater, batch_size=1)
class RoIHeads(torch.nn.Module):
__annotations__ = {
'box_coder': det_utils.BoxCoder,
'proposal_matcher': det_utils.Matcher,
cfg,
thr=0.1)
if im2show.shape[0] > 1100:
im2show = cv2.resize(im2show,
(int(1000. * float(im2show.shape[1]) / im2show.shape[0]), 1000)) # noqa
cv2.imshow('test', im2show)
cv2.waitKey(0)
with open(det_file, 'wb') as f:
pickle.dump(all_boxes, f, pickle.HIGHEST_PROTOCOL)
print('Evaluating detections')
imdb.evaluate_detections(all_boxes, output_dir)
if __name__ == '__main__':
# data loader
imdb = VOCDataset(imdb_name, cfg.DATA_DIR, cfg.batch_size,
yolo_utils.preprocess_test,
processes=1, shuffle=False, dst_size=cfg.multi_scale_inp_size)
net = Darknet19()
net_utils.load_net(trained_model, net)
net.cuda()
net.eval()
test_net(net, imdb, max_per_image, thresh, vis)
imdb.close()