def FasterRcnn_eval(dataset_path, ckpt_path, ann_file):
"""FasterRcnn evaluation."""
ds = create_fasterrcnn_dataset(dataset_path, batch_size=config.test_batch_size, is_training=False)
net = Faster_Rcnn_Resnet50(config)
param_dict = load_checkpoint(ckpt_path)
load_param_into_net(net, param_dict)
net.set_train(False)
eval_iter = 0
total = ds.get_dataset_size()
outputs = []
dataset_coco = COCO(ann_file)
print("\n========================================\n")
print("total images num: ", total)
print("Processing, please wait a moment.")
max_num = 128
for data in ds.create_dict_iterator(num_epochs=1):
eval_iter = eval_iter + 1
img_data = data['image']
img_metas = data['image_shape']
gt_bboxes = data['box']
gt_labels = data['label']
gt_num = data['valid_num']
start = time.time()
# run net
output = net(img_data, img_metas, gt_bboxes, gt_labels, gt_num)
end = time.time()
print("Iter {} cost time {}".format(eval_iter, end - start))
# output
all_bbox = output[0]
all_label = output[1]
all_mask = output[2]
for j in range(config.test_batch_size):
all_bbox_squee = np.squeeze(all_bbox.asnumpy()[j, :, :])
all_label_squee = np.squeeze(all_label.asnumpy()[j, :, :])
all_mask_squee = np.squeeze(all_mask.asnumpy()[j, :, :])
all_bboxes_tmp_mask = all_bbox_squee[all_mask_squee, :]
all_labels_tmp_mask = all_label_squee[all_mask_squee]
if all_bboxes_tmp_mask.shape[0] > max_num:
inds = np.argsort(-all_bboxes_tmp_mask[:, -1])
inds = inds[:max_num]
all_bboxes_tmp_mask = all_bboxes_tmp_mask[inds]
all_labels_tmp_mask = all_labels_tmp_mask[inds]
outputs_tmp = bbox2result_1image(all_bboxes_tmp_mask, all_labels_tmp_mask, config.num_classes)
outputs.append(outputs_tmp)
eval_types = ["bbox"]
result_files = results2json(dataset_coco, outputs, "./results.pkl")
coco_eval(result_files, eval_types, dataset_coco, single_result=True)
print("Create Mindrecord. It may take some time.")
data_to_mindrecord_byte_image("other", True, prefix)
print("Create Mindrecord Done, at {}".format(mindrecord_dir))
else:
print("image_dir or anno_path not exits.")
while not os.path.exists(mindrecord_file + ".db"):
time.sleep(5)
print("CHECKING MINDRECORD FILES DONE!")
loss_scale = float(config.loss_scale)
# When create MindDataset, using the fitst mindrecord file, such as FasterRcnn.mindrecord0.
dataset = create_fasterrcnn_dataset(mindrecord_file,
batch_size=config.batch_size,
device_num=device_num,
rank_id=rank)
dataset_size = dataset.get_dataset_size()
print("Create dataset done!")
net = Faster_Rcnn_Resnet50(config=config)
net = net.set_train()
load_path = args_opt.pre_trained
if load_path != "":
param_dict = load_checkpoint(load_path)
key_mapping = {
'down_sample_layer.1.beta':
'bn_down_sample.beta',
pre_trained = '/ckpt_path'
print("CHECKING MINDRECORD FILES ...")
if not os.path.exists(mindrecord_file):
if not os.path.isdir(mindrecord_dir):
os.makedirs(mindrecord_dir)
if os.path.isdir(config.coco_root):
print("Create Mindrecord. It may take some time.")
data_to_mindrecord_byte_image("coco", False, prefix, file_num=1)
print("Create Mindrecord Done, at {}".format(mindrecord_dir))
else:
print("coco_root not exits.")
print('Start generate adversarial samples.')
# build network and dataset
ds = create_fasterrcnn_dataset(mindrecord_file, batch_size=config.test_batch_size, \
repeat_num=1, is_training=False)
net = Faster_Rcnn_Resnet50(config)
param_dict = load_checkpoint(pre_trained)
load_param_into_net(net, param_dict)
net = net.set_train(False)
# build attacker
model = ModelToBeAttacked(net)
attack = GeneticAttack(model, model_type='detection', max_steps=50, reserve_ratio=0.3, mutation_rate=0.05,
per_bounds=0.5, step_size=0.25, temp=0.1)
# generate adversarial samples
sample_num = 5
ori_imagess = []
adv_imgs = []
ori_meta = []
