If a pre-trained model nessary?

[YMkai]

🚀 If a pre-trained model nessary?

I changed the cfg and the model structure,so now I have no pre-trained model to use ,regards of this situation,I trianed the model directly but it occurred some errors which showed below ,my question is that can I train the model without pre-trained model ？Thank you for your help.

I used this:
python train.py --data data/spot.data --cfg cfg/yolov3.cfg --epochs 10

Here is an error.

Namespace(accumulate=4, adam=False, arc='default', batch_size=16, bucket='', cache_images=False, cfg='cfg/yolov3.cfg', data='data/spot.data', device='', epochs=10, evolve=False, img_size=416, img_weights=False, multi_scale=False, name='', nosave=False, notest=False, prebias=False, rect=False, resume=False, transfer=False, var=None, weights='weights/ultralytics68.pt')
Using CPU

/home/nb505/.local/lib/python3.6/site-packages/tensorflow/python/framework/dtypes.py:523: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.
_np_qint8 = np.dtype([("qint8", np.int8, 1)])
/home/nb505/.local/lib/python3.6/site-packages/tensorflow/python/framework/dtypes.py:524: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.
_np_quint8 = np.dtype([("quint8", np.uint8, 1)])
/home/nb505/.local/lib/python3.6/site-packages/tensorflow/python/framework/dtypes.py:525: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.
_np_qint16 = np.dtype([("qint16", np.int16, 1)])
/home/nb505/.local/lib/python3.6/site-packages/tensorflow/python/framework/dtypes.py:526: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.
_np_quint16 = np.dtype([("quint16", np.uint16, 1)])
/home/nb505/.local/lib/python3.6/site-packages/tensorflow/python/framework/dtypes.py:527: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.
_np_qint32 = np.dtype([("qint32", np.int32, 1)])
/home/nb505/.local/lib/python3.6/site-packages/tensorflow/python/framework/dtypes.py:532: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.
np_resource = np.dtype([("resource", np.ubyte, 1)])
Traceback (most recent call last):
File "train.py", line 115, in train
chkpt['model'] = {k: v for k, v in chkpt['model'].items() if model.state_dict()[k].numel() == v.numel()}
File "train.py", line 115, in
chkpt['model'] = {k: v for k, v in chkpt['model'].items() if model.state_dict()[k].numel() == v.numel()}
KeyError: 'module_list.85.Conv2d.weight'

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
File "train.py", line 463, in
train() # train normally
File "train.py", line 120, in train
raise KeyError(s) from e
KeyError: "weights/ultralytics68.pt is not compatible with cfg/yolov3.cfg. Specify --weights '' or specify a --cfg compatible with weights/ultralytics68.pt. See #657"

Thank you very much for helping me!

[developer0hye]

python3 train.py --weights '' --cfg cfg/*.cfg # any cfg will work here

Try this one!

[glenn-jocher]

@mingkaaaaaai as the error message already states, use --weights ''. This will load no weights.