Comments (7)
I am a foreigner , I can't read japanese, I googled a lot of chainer example, But most of them is written in japanese, I want to know how to write a evaluator extension?
from chainercv.
Well, I don't know much example other than ones in ChainerCV that modify Evaluator.
observation = {}
with reporter.report_scope(observation):
reporter.report(report, target)
return observation
This basically does observation.update(report)
.
By looking at the source code of the original Evaluator
, you can find that our evaluator simply modified parts of it.
Actually, while trying to come up with an answer to your question, I realized that it can be simplified to return report
!
As a side note, the context with reporter.report_score(observation)
is useful when reporter.report
is called inside of a function in the with block.
Here is an example.
import chainer
from chainer import reporter
def f():
reporter.report({'a': 3})
if __name__ == '__main__':
observation = {}
r = reporter.Reporter()
with r:
with reporter.report_scope(observation):
f()
print(observation)
EDIT:
Actually, I had to do reporter.report
.
The with reporter.report_scope
is needed to get the namespace of reported values (See Evaluator.__call__
, where namespace is prepared).
from chainercv.
I just can't find where did it reported ?? in which log file path? I don't see any report F1 score or Accuracy in console print?
from chainercv.
If you use LogReport
in your train loop, the values will be printed to a file.
By default, the log file is result/log
.
(https://github.com/chainer/chainercv/blob/master/examples/faster_rcnn/train.py#L100)
from chainercv.
@yuyu2172 but you also use PrintReport
trainer.extend(extensions.PrintReport(
| ['iteration', 'epoch', 'elapsed_time', 'lr',
| 'main/loss',
| 'main/roi_loc_loss',
| 'main/roi_cls_loss',
| 'main/rpn_loc_loss',
| 'main/rpn_cls_loss',
| 'validation/main/map',
| ]), trigger=print_interval)
```
Does this 2 line conflict with each other,
I did not found any report log in my linux filesystem. I am try to find.
from chainercv.
mAP
gets recorded on the log file only when it gets evaluated.
You will find it in the log file at iteration 5000.
They don't conflict with each other.
from chainercv.
If you have more question, please reopen.
from chainercv.
Related Issues (20)
- Faster RCNN training result problem HOT 2
- Add a img.resize function in utils HOT 2
- A function to return segmented image HOT 2
- no module named 'chainercv.datasets' HOT 6
- Problems of FCIS HOT 6
- Problem about eval_detection HOT 2
- Accuracy problems of FCIS example HOT 5
- loc_normalize_std in ProposalTargetCreator HOT 5
- yolo/train_v3.py does not work HOT 2
- DirectoryParsingLabelDataset fails to read images with an alpha channel
- Allow empty object bounding box for SSD training
- `neg_iou_thresh_lo` value in `ProposalTargetCreator`
- Is it fixed for loading the trained weights for FPN model? HOT 2
- Change Request in chainercv/examples/fpn/train_multi.py HOT 1
- build wheels for chainerCV failed HOT 1
- can't install environment, invalid channel HOT 3
- "Introduction to Chainer" doc link broken
- Request for train.py for YOLO
- eval_semantic_segmentation and calc_semantic_segmentation_confusion for when we have ignore label
- possible bug in the way that mIoU is computed
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 chainercv.