Comments (15)
请问这个效果是crf后的效果吗? 论文里面的结果都是加了crf的效果。
from l2g.
谢谢您的回复!在以上结果中我加了crf之后,miou分别是vgg16_v1:68.5%,vgg16_v2,:69.3%,resnet_v1:70.4%,resnet_v2:71.3%,与论文的结果还是有小差别,比如说vgg16的结果比论文高,而resnet101的结果比论文低,我知道训练环境的不同也可能会导致这样的差别,所以我还想请问下,论文中的值是单次跑出来的结果还是多次跑出来的结果然后取平均呢?以及您在生成训练deeplab用的伪掩码时是否使用了crf?再次感谢!
from l2g.
论文是单次跑出来的结果,伪掩码没有加crf。你的伪标签在训练集上面的mIoU是多少? ResNet-based deeplab 是不是用的coco pretrained model 做的初始化?
from l2g.
谢谢,我在运行gen_gt.py时,使用的背景阈值是0.25,这时候生成的pseudo_seg_labels在train集上的mIoU是71.74%,在train_aug集上是69.93%,ResNet-based deeplab使用预训练模型是遵照readme下载的./pretrained/deeplabv1_resnet101-coco.pth,VGG-based deeplab分别是v1:./pretrained/vgg16_deeplabv1_pretrain.pth,v2:./pretrained/vgg16_pretrain.pth
from l2g.
方便提供下config文件吗?
from l2g.
好的谢谢,这是voc12_resnet_v1
EXP:
ID: voc12_resnet_v1
OUTPUT_DIR: data
DATASET:
NAME: vocaug
ROOT: /media/jnu/dgs/VOC2012/
LABELS: ./data/datasets/voc12/labels.txt
N_CLASSES: 21
IGNORE_LABEL: 255
SCALES: [0.5, 0.75, 1.0, 1.25, 1.5]
SPLIT:
TRAIN: train_aug
VAL: val
TEST: test
DATALOADER:
NUM_WORKERS: 0
IMAGE:
MEAN:
R: 122.675
G: 116.669
B: 104.008
SIZE:
BASE: # None
TRAIN: 321
TEST: 513
MODEL:
NAME: DeepLabV1_ResNet101_MSC
N_BLOCKS: [3, 4, 23, 3]
ATROUS_RATES: [6, 12, 18, 24]
INIT_MODEL: ./pretrained/deeplabv1_resnet101-coco.pth
SOLVER:
BATCH_SIZE:
TRAIN: 5
TEST: 1
ITER_MAX: 20000
ITER_SIZE: 2
ITER_SAVE: 5000
ITER_TB: 100
LR_DECAY: 10
LR: 2.5e-4
MOMENTUM: 0.9
OPTIMIZER: sgd
POLY_POWER: 0.9
WEIGHT_DECAY: 5.0e-4
AVERAGE_LOSS: 20
CRF:
ITER_MAX: 10
POS_W: 3
POS_XY_STD: 3
BI_W: 4
BI_XY_STD: 121
BI_RGB_STD: 5
这是 voc12_resnet_v2
EXP:
ID: voc12_resnet_v2
OUTPUT_DIR: data
DATASET:
NAME: vocaug
ROOT: /media/jnu/dgs/VOC2012/
LABELS: ./data/datasets/voc12/labels.txt
N_CLASSES: 21
IGNORE_LABEL: 255
SCALES: [0.5, 0.75, 1.0, 1.25, 1.5]
SPLIT:
TRAIN: train_aug
VAL: val
TEST: test
DATALOADER:
NUM_WORKERS: 0
IMAGE:
MEAN:
R: 122.675
G: 116.669
B: 104.008
SIZE:
BASE: # None
TRAIN: 321
TEST: 513
MODEL:
NAME: DeepLabV2_ResNet101_MSC
N_BLOCKS: [3, 4, 23, 3]
ATROUS_RATES: [6, 12, 18, 24]
INIT_MODEL: ./pretrained/deeplabv1_resnet101-coco.pth
SOLVER:
BATCH_SIZE:
TRAIN: 5
TEST: 1
ITER_MAX: 20000
ITER_SIZE: 2
ITER_SAVE: 5000
ITER_TB: 100
LR_DECAY: 10
LR: 2.5e-4
MOMENTUM: 0.9
OPTIMIZER: sgd
POLY_POWER: 0.9
WEIGHT_DECAY: 5.0e-4
AVERAGE_LOSS: 20
CRF:
ITER_MAX: 10
POS_W: 3
POS_XY_STD: 1
BI_W: 4
BI_XY_STD: 67
BI_RGB_STD: 3
from l2g.
我又训练了一遍,resnet_v1:70%,resnet_v2:71.4%,和第一次训练的结果区别不是很大,请问能否提供您生成的用于训练deeplab的伪掩码,谢谢
from l2g.
你好,我们生成的伪标签就是用提供的预训练分类模型生成的,你可以直接生成下。 此外,可否提供下你的预训练的分割模型,我们确认下问题是出在测试阶段还是训练阶段?
from l2g.
不好意思我没有理解,“伪标签就是用提供的预训练分类模型生成的”,我没有找到预训练分类模型,请问在哪里能下载呢,还有“预训练的分割模型”是指deeplabv1_resnet101-coco.pth这个文件吗
from l2g.
抱歉没有说明清楚,“提供的预训练模型”是指这个,你可以尝试一下。
“预训练的分割模型”是指你训练后得到的deeplabv1和deeplabv2模型,我们希望在我们这里测试一下模型从而确定测试过程没有问题。
from l2g.
这是我训练后得到的deeplabv1_resnet101和deeplabv2_resnet101模型,麻烦您了。链接:https://pan.baidu.com/s/1cju8ilH1ghPo0mWfyZ4_jw
提取码:l2x3
from l2g.
@DL3399 hello, 请问你用我们的伪标签训练过模型了吗?
from l2g.
嗯嗯,我训练过了,在val集上结果分别是vgg16_v1:68.7%,vgg16_v2,:69.7%,resnet_v1:70.5%,resnet_v2:71.5%
from l2g.
使用caffe转化的预训练模型可以得到论文相应的结果,同时也好于pytorch模型的初始化。
这个(提取码48kv)是caffe转化的预训练模型,你可以试一试。
from l2g.
谢谢!
from l2g.
Related Issues (20)
- PASCAL VOC2012测试集的性能 HOT 1
- 预测测试集出来是空的 HOT 1
- resnet模型 HOT 2
- 怎么用自己的数据集? HOT 1
- 显著性图的问题
- request for initial weights of classification HOT 1
- crop_size的问题 HOT 1
- bg_threshold问题 HOT 5
- letter of thanks HOT 1
- 关于epoch HOT 2
- COCO用deeplabv2_resnet的配置文件训练,测试结果出来除了背景被分割,其他类miou值都是0 HOT 7
- AttributeError: 'Net' object has no attribute 'module' HOT 1
- Discrepancy in Pseudo Label Accuracy & Repository Settings for COCO
- 网络结构 HOT 1
- densecrf HOT 1
- 。
- pascal voc测试结果问题 HOT 2
- 训练二分类 HOT 1
- 流程存在问题,按readme第二阶段分割流程没有使用伪标签 HOT 2
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from l2g.