Comments (6)
Smoothing97
commented on July 3, 2024
在tensorboard里看训练到44460step输出的final_detection_img,感觉还可以？但是ckpt检测出来就很差，两者反差太大了，所以训练过程中输出的final_detection_img可以作为参考吗？
from nas_fpn_tensorflow.
yangxue0827
commented on July 3, 2024
可能是你在测试的时候没有加载上训练好的权重，加载的是预训练的 @Smoothing97
from nas_fpn_tensorflow.
Smoothing97
commented on July 3, 2024
@yangxue0827 不，我特意输出了ckpt的路径，
checkpoint_path=============> /home/smoothing/NAS_FPN_Tensorflow-master/output/trained_weights/FPN_resnet50_v1d_DOTA_20190515_nas_2_Part5/dota_32302model.ckpt
model restore from : /home/smoothing/NAS_FPN_Tensorflow-master/output/trained_weights/FPN_resnet50_v1d_DOTA_20190515_nas_2_Part5/dota_32302model.ckpt
from nas_fpn_tensorflow.
Smoothing97
commented on July 3, 2024
我从55000张800大小的水平数据里随机抽了10000张，一个1poch是10000，学习率采用warmup_cosine，warm step我设置了前4个epoch，两块1080TI训练，所以在20000step的时候
学习率warmup到了我设定的0.01*2。
让我头疼的是，如果我的权重没加载错（从输出的路径来看是的），用demo.py检测每一张图，都是一个样子的检测框，置信度还都很低....
from nas_fpn_tensorflow.
Smoothing97
commented on July 3, 2024
@yangxue0827 ：）打开在这个10000张随机水平800大小的DOTA数据训练了34000step的tensorboard文件，发现训练阶段显示的final_detection_img效果也不怎么样呢【手动微笑
看了一下score_greater_05_rois....
嗯没看出什么来，所以又看了一下score_greater_09_rois....
ROI生成的这么清奇...按照算法里还要取score高的roi作为proposal_target_layer的输入...
fpn_box_pred这两个值应该是网络自己预测出来的，所以问题应该是在我设置的anchor参数有问题（吧？
from nas_fpn_tensorflow.
yangxue0827
commented on July 3, 2024
前期因为就加了wareup，所以你看前期tensorboard的效果不会那么好，需要多训练一会。另外建议你测一下训练集图片，如果没有结果那就是测试代码的问题了。 @Smoothing97
from nas_fpn_tensorflow.
Related Issues (20)
- NUM_FPN 这个参数什么含义 HOT 4
- NAS+Cascade loss 无法下降 HOT 2
- Question about BN in nas_fpn
- Question on top-down layer in NAS-FPN
- ValueError: Dimensions must be equal, but are 256 and 384 for 'tower_0/build_pyramid/build_P4/add' (op: 'Add') with input shapes: [1,?,?,256], [1,?,?,384]. HOT 3
- where is the RNN controller? HOT 1
- Closed because of impoliteness
- Is the code really slow?
- Questions
- Would you please provide AP result on COCO 2017 test-dev?
- ModuleNotFoundError: No module named 'tensorflow.contrib' HOT 1
- mAP question
- tensorflow contrib' module not found
- tensorflow contrib' module not found
- Link failure HOT 1
- how to train vocdata?
- GPU is not utilized and no progress happening
- resnet101_v1d model load error
- How to train, to do object detection for single class using NAS-FPN? HOT 1
- Can this be used to train on DOTA dataset ?
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 nas_fpn_tensorflow.