-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathcloud_hmdb51_efficientnetb3_branch.yml
More file actions
91 lines (78 loc) · 2.44 KB
/
cloud_hmdb51_efficientnetb3_branch.yml
File metadata and controls
91 lines (78 loc) · 2.44 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
help: False
show_model: False
base_path: /HOME/scz0831/run/3d/TLEE/ckpt
train_mode: test # choice = [1, 1.1, 2, test]
test_path: exp1_hmdb51_mp_efficientnetb3/0.2_mp_gm # 文件夹名即可 训练模式2, test都需要
b: '0.2'
# checkpoint
resume: False
resume_path:
seed: 2
##################
data:
name: hmdb51
path_frame: /HOME/scz0831/run/3d/HMDB51/jpg/ # set the dataset dir
path_split: /HOME/scz0831/run/3d/HMDB51/file_list/hmdb51_rgb_train_split_1.txt
path_label: /HOME/scz0831/run/3d/HMDB51/file_list/hmdb51_rgb_train_split_1.txt
path_classid: /HOME/scz0831/run/3d/HMDB51/labels/classInd.txt
preprocessing:
to_tensor: True
resize: 224
crop_center: 224
normalize:
mean: [0.43216, 0.394666, 0.37645]
std: [0.22803, 0.22145, 0.216989]
preprocessing_label:
video_multihot_labels: ""
one_hot_encoding: 51
num_frames: 10
num_workers: 6
shuffle: False
data_val:
name: hmdb51
path_frame: /HOME/scz0831/run/3d/HMDB51/jpg/ # set the dataset dir
path_split: /HOME/scz0831/run/3d/HMDB51/file_list/hmdb51_rgb_val_split_1.txt
path_label: /HOME/scz0831/run/3d/HMDB51/file_list/hmdb51_rgb_val_split_1.txt
path_classid: /HOME/scz0831/run/3d/HMDB51/labels/classInd.txt
preprocessing:
to_tensor: True
resize: 224
crop_center: 224
normalize:
mean: [0.43216, 0.394666, 0.37645]
std: [0.22803, 0.22145, 0.216989]
preprocessing_label:
video_multihot_labels: ""
one_hot_encoding: 51
num_frames: 10
num_workers: 6
shuffle: False
###################
model:
type: ts # [ts, t]
epochs: 10
useLRSchedule: True
schedule: [10, 20] # not enable if useLRSchedule is False
# branch settings
name: efficientnetb3
output_dim: [1536] # branches [3,] # 2048: ResNet50 512: ResNet18, ResNet34
branch_num: 2
branch_scale: 3 # the channel reduction ratio
crm_num: 3 # #crm
exit_thresh: 1
useFeatureFunc: mp # [frm, afrm, ema, mp, ap, na, none]
useGateFunc: gm # [eg, gm, random]
first_threshold: 0.99
single_frame_mac: [0.045, 0.295, 0.629, 0.992]
backbone:
type: 2d
pretrained: True
name: a.a
requires_grad: True
truncate_modules: -1 # index indicates up until which layer to keep modules, i.e. -1 removes the classifier.
pooling:
name: max
input_order8: [3, 6, 7, 0, 4, 5, 2, 1] #n8
input_order10: [5, 0, 9, 2, 7, 4, 6, 3, 8, 1] #n10
input_order16: [9, 8, 6, 3, 15, 1, 0, 12, 14, 2, 10, 7, 13, 11, 4, 5] # n16
num_classes: 51