Extract seed and cuda initialization utils

detect.py

@@ -5,8 +5,6 @@ from models import *
 from utils.datasets import *
 from utils.utils import *
 
-cuda = torch.cuda.is_available()
-device = torch.device('cuda:0' if cuda else 'cpu')
 f_path = os.path.dirname(os.path.realpath(__file__)) + '/'
 
 parser = argparse.ArgumentParser()
@@ -28,6 +26,10 @@ print(opt)
 
 
 def main(opt):
+
+    device = torch_utils.select_device()
+    print("Using device: \"{}\"".format(device))
+
     os.system('rm -rf ' + opt.output_folder)
     os.makedirs(opt.output_folder, exist_ok=True)
 

test.py

@@ -4,6 +4,8 @@ from models import *
 from utils.datasets import *
 from utils.utils import *
 
+from utils import torch_utils
+
 parser = argparse.ArgumentParser(prog='test.py')
 parser.add_argument('-batch_size', type=int, default=32, help='size of each image batch')
 parser.add_argument('-cfg', type=str, default='cfg/yolov3.cfg', help='path to model config file')
@@ -18,11 +20,11 @@ parser.add_argument('-img_size', type=int, default=416, help='size of each image
 opt = parser.parse_args()
 print(opt, end='\n\n')
 
-cuda = torch.cuda.is_available()
-device = torch.device('cuda:0' if cuda else 'cpu')
-
 
 def main(opt):
+    device = torch_utils.select_device()
+    print("Using device: \"{}\"".format(device))
+
     # Configure run
     data_config = parse_data_config(opt.data_config_path)
     nC = int(data_config['classes'])  # number of classes (80 for COCO)
@@ -128,4 +130,7 @@ def main(opt):
 
 
 if __name__ == '__main__':
+
+    init_seeds()
+
     mAP = main(opt)

train.py

@@ -6,6 +6,8 @@ from models import *
 from utils.datasets import *
 from utils.utils import *
 
+from utils import torch_utils
+
 parser = argparse.ArgumentParser()
 parser.add_argument('-epochs', type=int, default=100, help='number of epochs')
 parser.add_argument('-batch_size', type=int, default=16, help='size of each image batch')
@@ -26,20 +28,15 @@ print(opt)
 sys.argv[1:] = []  # delete any train.py command-line arguments before they reach test.py
 import test  # must follow sys.argv[1:] = []
 
-cuda = torch.cuda.is_available()
-device = torch.device('cuda:0' if cuda else 'cpu')
 
-random.seed(0)
+def main(opt):
-np.random.seed(0)
+
-torch.manual_seed(0)
+    device = torch_utils.select_device()
-if cuda:
+    print("Using device: \"{}\"".format(device))
-    torch.cuda.manual_seed(0)
+
-    torch.cuda.manual_seed_all(0)
     if not opt.multi_scale:
         torch.backends.cudnn.benchmark = True
 
-
-def main(opt):
     os.makedirs('weights', exist_ok=True)
 
     # Configure run
@@ -217,5 +214,8 @@ def main(opt):
 
 
 if __name__ == '__main__':
+
+    init_seeds()
+
     torch.cuda.empty_cache()
     main(opt)

utils/torch_utils.py

@@ -0,0 +1,23 @@
+import torch
+
+
+def check_cuda():
+    return torch.cuda.is_available()
+
+
+CUDA_AVAILABLE = check_cuda()
+
+
+def init_seeds(seed=0):
+    torch.manual_seed(seed)
+    if CUDA_AVAILABLE:
+        torch.cuda.manual_seed(seed)
+        torch.cuda.manual_seed_all(seed)
+
+
+def select_device(force_cpu=False):
+    if force_cpu:
+        device = torch.device('cpu')
+    else:
+        device = torch.device('cuda:0' if CUDA_AVAILABLE else 'cpu')
+    return device

utils/utils.py

@@ -5,11 +5,19 @@ import numpy as np
 import torch
 import torch.nn.functional as F
 
+from utils import torch_utils
+
 # Set printoptions
 torch.set_printoptions(linewidth=1320, precision=5, profile='long')
 np.set_printoptions(linewidth=320, formatter={'float_kind': '{:11.5g}'.format})  # format short g, %precision=5
 
 
+def init_seeds(seed=0):
+    random.seed(seed)
+    np.random.seed(seed)
+    torch_utils.init_seeds(seed=seed)
+
+
 def load_classes(path):
     """
     Loads class labels at 'path'