-
Notifications
You must be signed in to change notification settings - Fork 13
/
config.py
85 lines (80 loc) · 3.55 KB
/
config.py
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
# @Time : 2020/7/8
# @Author : Lart Pang
# @Email : lartpang@163.com
# @File : config.py
# @Project : HDFNet
# @GitHub : https://github.com/lartpang
import os
__all__ = ["proj_root", "arg_config"]
proj_root = os.path.dirname(__file__)
datasets_root = "/home/lart/Datasets/"
msra10k_path = os.path.join(datasets_root, "Saliency/RGBSOD", "MSRA10K")
ecssd_path = os.path.join(datasets_root, "Saliency/RGBSOD", "ECSSD")
dutomron_path = os.path.join(datasets_root, "Saliency/RGBSOD", "DUT-OMRON")
hkuis_path = os.path.join(datasets_root, "Saliency/RGBSOD", "HKU-IS")
pascals_path = os.path.join(datasets_root, "Saliency/RGBSOD", "PASCAL-S")
dutstr_path = os.path.join(datasets_root, "Saliency/RGBSOD", "DUTS/Train")
dutste_path = os.path.join(datasets_root, "Saliency/RGBSOD", "DUTS/Test")
lfsd_path = os.path.join(datasets_root, "Saliency/RGBDSOD", "LFSD")
rgbd135_path = os.path.join(datasets_root, "Saliency/RGBDSOD", "RGBD135")
dutrgbdtr_path = os.path.join(datasets_root, "Saliency/RGBDSOD", "DUT-RGBD/Train")
dutrgbdte_path = os.path.join(datasets_root, "Saliency/RGBDSOD", "DUT-RGBD/Test")
sip_path = os.path.join(datasets_root, "Saliency/RGBDSOD", "SIP")
ssd_path = os.path.join(datasets_root, "Saliency/RGBDSOD", "SSD")
stereo797_path = os.path.join(datasets_root, "Saliency/RGBDSOD", "STEREO-797")
stereo1000_path = os.path.join(datasets_root, "Saliency/RGBDSOD", "STEREO-1000")
rgbdtr_path = os.path.join(proj_root, "utils/data/data_list", "rgbd_train_jw.lst")
njudte_path = os.path.join(proj_root, "utils/data/data_list", "njud_test_jw.lst")
nlprte_path = os.path.join(proj_root, "utils/data/data_list", "nlpr_test_jw.lst")
# 配置区域 #####################################################################
arg_config = {
# 常用配置
"model": "HDFNet_VGG19",
"suffix": "7Datasets",
"resume": True, # 是否需要恢复模型
"use_aux_loss": True, # 是否使用辅助损失
"save_pre": True, # 是否保留最终的预测结果
"epoch_num": 30, # 训练周期
"lr": 0.005,
"data_mode": "RGBD", # 'RGB'/'RGBD' 任务模式，支持RGB和RGBD两种类型任务的训练与测试
# RGBD
"rgbd_data": {
"tr_data_path": rgbdtr_path,
# "tr_data_path": dutrgbdtr_path,
"te_data_list": {
# "dutrgbd": dutrgbdte_path,
"lfsd": lfsd_path,
"njud": njudte_path,
"nlpr": nlprte_path,
"rgbd135": rgbd135_path,
"sip": sip_path,
"ssd": ssd_path,
"stereo797": stereo797_path,
"stereo1000": stereo1000_path,
},
},
# RGB
"rgb_data": {
"tr_data_path": dutstr_path,
"te_data_list": {
"dutomron": dutomron_path,
"hkuis": hkuis_path,
"ecssd": ecssd_path,
"pascals": pascals_path,
"duts": dutste_path,
},
},
"print_freq": 10, # >0, 保存迭代过程中的信息
"prefix": (".jpg", ".png"),
# img_prefix, gt_prefix，用在使用索引文件的时候的对应的扩展名，这里使用的索引文件不包含后缀
"reduction": "mean", # 损失处理的方式，可选“mean”和“sum”
"optim": "sgd_trick", # 自定义部分的学习率
"weight_decay": 5e-4, # 微调时设置为0.0001
"momentum": 0.9,
"nesterov": False,
"lr_type": "poly",
"lr_decay": 0.9, # poly
"batch_size": 4, # 要是继续训练, 最好使用相同的batchsize
"num_workers": 8, # 不要太大, 不然运行多个程序同时训练的时候, 会造成数据读入速度受影响
"input_size": 320,
}