ithron / sof-utils Goto Github PK
View Code? Open in Web Editor NEWUtilities to aid working with the SOF dataset
License: BSD 3-Clause "New" or "Revised" License
Utilities to aid working with the SOF dataset
License: BSD 3-Clause "New" or "Revised" License
Is your feature request related to a problem? Please describe.
For nearly all supervised task, only the radiographs from the first visit should be used. Therefore, it would be more convenient to add a visit selection option to the sot-export-images
tool than to manually delete unwanted visits after exporting.
Describe the solution you'd like
An option to select the visits to be exported should be added to sof-export-images
.
The option should accept a list of visits (integer) that should be exported.
If not visit is given, all visits should be exported.
Describe alternatives you've considered
Export all visits and manually delete the unwanted visits afterwards.
However, this approach is suboptimal, since it requires exporting more images than required, resulting in longer run-times and disk space usage.
Is your feature request related to a problem? Please describe.
When exporting images using sof-export-images
eg. for annotation training data, only IDs in the training set should be included, or the other way around, the IDs from the test set should be excluded.
Describe the solution you'd like
Add an option to sof-export-images
to include or exclude IDs listed in a file.
Describe alternatives you've considered
Manual deletion of unwanted files after the export, but costs more time and disk space.
Describe the bug
There is at least one upside down radiograph in the dataset where sof-detect-keypoints
labels both sides a 'Complete', one side with a high score and one side with a very low score and incorrect key point positions.
In this case the tool does not detect that the radiograph is upside down.
To Reproduce
Detect key points on SOF ID 41886 visit 1.
Is your feature request related to a problem? Please describe.
When images are exported using sof-export-images
for annotation, the exported images must be grouped manually in order to distribute them to multiple annotators.
Manual randomized grouping, e.g. to split centers across annotators, is very difficult.
Describe the solution you'd like
There should be an option to sof-export-images
to automatically, and optionally randomly, group the annotated images, e.g. by adding a group prefix to the file names or by exporting to different directories.
Describe alternatives you've considered
Is your feature request related to a problem? Please describe.
Exporting images using sof-export-images
is fine. But in order to use them in tools like LabelStudio, they must be contained in a ZIP archive. For a single group this can be easily done manually, but for grouped export an automated method would be preferable.
Describe the solution you'd like
Add option to sof-export-images
to write the files directly to a ZIP archive, or to multiple archives is the grouping case.
Describe alternatives you've considered
Manual zipping is not an alternative.
The hip images can be split trivially into images of the left and right hip.
Is your feature request related to a problem? Please describe.
The TFDS is easy to use, by the TF ObjectDetection API requires a simple .tfrecord file with a slightly different structure.
Describe the solution you'd like
Add a conversion tool to the package to convert between the SOF_hip TFDS dataset and the format required by TF ObjectDetection API.
A declarative, efficient, and flexible JavaScript library for building user interfaces.
๐ Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. ๐๐๐
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google โค๏ธ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.