Comments (9)
可视化比较简单,这样做
`,anchors,=generate_default_anchor_maps()
print type(anchors),anchors.shape
print len(anchors)
for anc in anchors:
if anc[0]<0 or anc[1]<0 or anc[2]<0 or anc[3]<0:
print anc
import matplotlib.pyplot as plt
xmin=-224
xmax=672
ymin=-224
ymax=672
fig = plt.figure(figsize=(8,8))
ax = fig.add_subplot(111)
plt.xlim( (xmin, xmax) )
plt.ylim( (ymin, ymax) )
#rect = plt.Rectangle((0.1,0.1),0.5,0.3)
#ax.add_patch(rect)
for anch in anchors:
rect = plt.Rectangle((anch[1],anch[0]),#(x0,y0)
anch[3]-anch[1],#width
anch[2]-anch[0],#height
edgecolor='r',
facecolor='none')
ax.add_patch(rect)
plt.show()`
from nts-net.
I want to know how to draw the box when testing , so that i can find if the network focus the most import regins. The paper had a amazing visualization.
from nts-net.
Actually, I did it by my mentioned approach not zbxzc35 mentioned . But it seems does not have amazing visualization , which anchor boxes should be localized discriminative feature regions on image directly, my visualization result told me that effect of RPN module still confuses me. It need author's help to tell more details.
from nts-net.
@VCBE123 I think you need to find anchor boxes selected by trained model and pad your image with constant value for resizing image's size and then, the way of drawing anchor boxes on PIL image format is easy on searching from stackoverflow.com, good luck for you, guy.
from nts-net.
@zhouyuangan Thank you for your help.
from nts-net.
@zbxzc35 为什么坐标是 -224 672 输入图片大小不是 448*448 吗,这样的设置不会导致坐标超出图片的大小吗
from nts-net.
@VCBE123 I think you need to find anchor boxes selected by trained model and pad your image with constant value for resizing image's size and then, the way of drawing anchor boxes on PIL image format is easy on searching from stackoverflow.com, good luck for you, guy.
Hello, when I run test.py on CUB_200 dataset, nothing has been output.Is this normal? then where to find the result of bounding boxes of the test images ? without the bbox information how can I visualize? I really need help.
from nts-net.
@VCBE123 I think you need to find anchor boxes selected by trained model and pad your image with constant value for resizing image's size and then, the way of drawing anchor boxes on PIL image format is easy on searching from stackoverflow.com, good luck for you, guy.
Hello, when I run test.py on CUB_200 dataset, nothing has been output.Is this normal? then where to find the result of bounding boxes of the test images ? without the bbox information how can I visualize? I really need help.
I would like to ask, did you successfully draw the bounding box during the test? How to visualize? thank you very much.
from nts-net.
from nts-net.
Related Issues (20)
- 谢谢作者的无私分享。我想自己制作数据集用此网络试试效果,请问我该怎么标注图片? HOT 6
- About the figure of training the model
- Typo error.
- My owndataset has a SyntaxError: not a TIFF file (header b'' not valid) HOT 1
- 依赖库的版本问题
- small batch_size, better result !
- Set the rationality of the partcls_loss HOT 1
- permissionError : [Errno 13 ] Permission denied : '/data_4t' HOT 1
- 对于代码中ProposalNet有些疑问
- When I trained NTS-Net on my own dataset and stanford dog dataset, the train accuracy can reach 99.8%, but the test accuracy just 85%. Do you same as me HOT 1
- This code version is sooo ~ bad! and memory error why???
- Different test accuracy when testing with different BATCH_SIZE in config.py HOT 5
- error in resnet
- 模型文件无法下载??
- NTS-Net for small image size
- 为什么要在输入的图像四周补0,这样不会导致截取的区域全0么
- what's the aim of padding 0? what's the meaning of cat_num? HOT 2
- ValueError: not enough values to unpack (expected 2, got 0) HOT 2
- test loss: 0.737 and test acc: 0.849 total sample: 5794
- 有人试过nts-net在小数据集的性能吗,我自定义数据集上面模型一直欠拟合underfitting,怎么回事呢
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 nts-net.