def initialize(self,layer):
if os.path.isabs(self.net_path):
net_path_full = self.net_path
else:
net_path_full = os.path.join(env_settings().network_path, self.net_path)
self.net, _ = load_network(net_path_full, backbone_pretrained=False)
if self.use_gpu:
self.net.cuda()
self.net.eval()
self.iou_predictor = self.net.bb_regressor
self.layer_stride = {'conv1': 2, 'layer1': 4, 'layer2': 8, 'layer3': 16, 'layer4': 32, 'classification': 16, 'fc': None}
self.layer_dim = {'conv1': 64, 'layer1': 64, 'layer2': 128, 'layer3': 256, 'layer4': 512, 'classification': 256,'fc': None}
self.iounet_feature_layers = self.net.bb_regressor_layer
if isinstance(self.pool_stride, int) and self.pool_stride == 1:
self.pool_stride = [1]*len(self.output_layers)
#print(self.output_layers)
self.output_layers = list(layer)
self.feature_layers = sorted(list(set(self.output_layers + self.iounet_feature_layers)))
#print('ajun',self.output_layers1,self.output_layers1,self.output_layers1 + self.iounet_feature_layers,self.output_layers + self.iounet_feature_layers)
self.mean = torch.Tensor([0.485, 0.456, 0.406]).view(1,-1,1,1)
self.std = torch.Tensor([0.229, 0.224, 0.225]).view(1,-1,1,1)
def initialize(self):
if os.path.isabs(self.net_path):
net_path_full = self.net_path
else:
net_path_full = os.path.join(env_settings().network_path,
self.net_path)
self.net, _ = load_network(net_path_full, backbone_pretrained=False)
self.net.cuda()
self.net.eval()
self.mean = torch.Tensor([0.485, 0.456, 0.406]).view(1, -1, 1,
1).cuda()
self.std = torch.Tensor([0.229, 0.224, 0.225]).view(1, -1, 1, 1).cuda()
self.layer_stride = {
'conv1': 2,
'layer1': 4,
'layer2': 8,
'layer3': 16,
'layer4': 32,
'classification': 16,
'fc': None
}
self.layer_dim = {
'conv1': 64,
'layer1': 64,
'layer2': 128,
'layer3': 256,
'layer4': 512,
'classification': 256,
'fc': None
}
self.iou_predictor = self.net.bb_regressor
self.location_predictor = self.net.location_predictor
#self.iounet_feature_layers = self.net.bb_regressor_layer
#self.locator_feature_layers = self.net.location_predictor_layer
self.iounet_feature_layers = ('layer2', 'layer3')
self.locator_feature_layers = ('layer2', 'layer3')
self.backbone_feature_layers = sorted(
list(set(self.iounet_feature_layers +
self.locator_feature_layers)))
## all parameters do not require grad
for p in self.net.parameters():
p.requires_grad = False
def initialize(self):
if os.path.isabs(self.net_path):
net_path_full = self.net_path
else:
net_path_full = os.path.join(env_settings().network_path,
self.net_path)
self.net, _ = load_network(net_path_full, backbone_pretrained=False)
self.net.cuda()
self.net.eval()
self.mean = torch.Tensor([0.485, 0.456, 0.406]).view(1, -1, 1,
1).cuda()
self.std = torch.Tensor([0.229, 0.224, 0.225]).view(1, -1, 1, 1).cuda()
self.iou_predictor = self.net.bb_regressor
self.iou_feature_extractor = self.net.bb_regressor.get_iou_feat
self.locator_feature_extractor = self.net.location_predictor.get_locator_feat
## all parameters do not require grad
for p in self.net.parameters():
p.requires_grad = False