Change the --img-size parameter into a int list to specify depth, height and width by a parameter

inference.py

@@ -31,8 +31,8 @@ parser.add_argument('-j', '--workers', default=2, type=int, metavar='N',
                     help='number of data loading workers (default: 2)')
 parser.add_argument('-b', '--batch-size', default=256, type=int,
                     metavar='N', help='mini-batch size (default: 256)')
-parser.add_argument('--img-size', default=None, type=int,
-                    metavar='N', help='Input image dimension')
+parser.add_argument('--img-size', default=None, nargs=3, type=int,
+                    metavar='N N N', help='Input image dimension (e.g. --img-size 3 224 224), uses model default if empty')
 parser.add_argument('--mean', type=float, nargs='+', default=None, metavar='MEAN',
                     help='Override mean pixel value of dataset')
 parser.add_argument('--std', type=float, nargs='+', default=None, metavar='STD',

timm/data/config.py

@@ -23,8 +23,10 @@ def resolve_data_config(args, default_cfg={}, model=None, verbose=True):
         input_size = tuple(args['input_size'])
         in_chans = input_size[0]  # input_size overrides in_chans
     elif 'img_size' in args and args['img_size'] is not None:
-        assert isinstance(args['img_size'], int)
-        input_size = (in_chans, args['img_size'], args['img_size'])
+        assert isinstance(args['img_size'], (tuple, list))
+        assert len(args['img_size']) == 3
+        input_size = tuple(args['img_size'])
+        in_chans = input_size[0]  # input_size overrides in_chans
     elif 'input_size' in default_cfg:
         input_size = default_cfg['input_size']
     new_config['input_size'] = input_size

train.py

@@ -80,8 +80,8 @@ parser.add_argument('--num-classes', type=int, default=1000, metavar='N',
                     help='number of label classes (default: 1000)')
 parser.add_argument('--gp', default=None, type=str, metavar='POOL',
                     help='Global pool type, one of (fast, avg, max, avgmax, avgmaxc). Model default if None.')
-parser.add_argument('--img-size', type=int, default=None, metavar='N',
-                    help='Image patch size (default: None => model default)')
+parser.add_argument('--img-size', default=None, nargs=3, type=int,
+                    metavar='N N N', help='Input image dimension (e.g. --img-size 3 224 224), uses model default if empty')
 parser.add_argument('--crop-pct', default=None, type=float,
                     metavar='N', help='Input image center crop percent (for validation only)')
 parser.add_argument('--mean', type=float, nargs='+', default=None, metavar='MEAN',

validate.py

@@ -50,8 +50,8 @@ parser.add_argument('-j', '--workers', default=4, type=int, metavar='N',
                     help='number of data loading workers (default: 2)')
 parser.add_argument('-b', '--batch-size', default=256, type=int,
                     metavar='N', help='mini-batch size (default: 256)')
-parser.add_argument('--img-size', default=None, type=int,
-                    metavar='N', help='Input image dimension, uses model default if empty')
+parser.add_argument('--img-size', default=None, nargs=3, type=int,
+                    metavar='N N N', help='Input image dimension (e.g. --img-size 3 224 224), uses model default if empty')
 parser.add_argument('--crop-pct', default=None, type=float,
                     metavar='N', help='Input image center crop pct')
 parser.add_argument('--mean', type=float, nargs='+', default=None, metavar='MEAN',