You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hi, i implemented the code to my dataset.
I have 360 video in total for 2 class (270 training 90 for validation).
I select the "sample duration" = 32
I didn't quite understand the batch size and epoch.
When i set batch_size = 12, training dataset for per epoch = 270/12 = 23 video
So, the batch size is, from 23 video, select 12 frame and train?
This selected 23 video, is it selected randomly per epoch from 270 video? or is it the same 23 video for full 200 epoch training?
Can anyone explain please? Thank you.
The opts.py i selected
parser.add_argument('--sample_size',
default=224,
type=int,
help='Height and width of inputs')
parser.add_argument('--sample_duration',
default=32,
type=int,
help='Temporal duration of inputs')
parser.add_argument('--batch_size',
default=12,
type=int,
help='Batch Size')
parser.add_argument('--n_val_samples',
default=12,
type=int,
help='Number of validation samples for each activity')
The text was updated successfully, but these errors were encountered:
Hi, i implemented the code to my dataset.
I have 360 video in total for 2 class (270 training 90 for validation).
I select the "sample duration" = 32
I didn't quite understand the batch size and epoch.
When i set batch_size = 12, training dataset for per epoch = 270/12 = 23 video
So, the batch size is, from 23 video, select 12 frame and train?
This selected 23 video, is it selected randomly per epoch from 270 video? or is it the same 23 video for full 200 epoch training?
Can anyone explain please? Thank you.
The opts.py i selected
parser.add_argument('--sample_size',
default=224,
type=int,
help='Height and width of inputs')
parser.add_argument('--sample_duration',
default=32,
type=int,
help='Temporal duration of inputs')
parser.add_argument('--batch_size',
default=12,
type=int,
help='Batch Size')
parser.add_argument('--n_val_samples',
default=12,
type=int,
help='Number of validation samples for each activity')
The text was updated successfully, but these errors were encountered: