Releases: CertifaiAI/classifai
V2 Alpha 3 Release, Different Label Colors, Add Data, Project Statistics
Classifai V2.0.0-alpha3
21 December 2021
What News
- Different colors for labels
Different colors are assigned for labels for better differentiating the labeled objects
- Add image to project folder tool
Users can add images from different source folder into the current project folder with ease
- Project Statistics
Users can check the number of labeled images, number of unlabeled images, and number of labels per class of the current project, this will help track the progress of annotation and have a good understanding of the class distribution
- Shortcut key to toggle tools
User can use short cut key to toggle pointer tool and annotation tool, this will speed up labelling process
- Toggle tabs
Open and close the required tabs easily by clicking the icons at the right side tab
- Shift canvas in zoom mode
Users are now able to shift the image to navigate to the area of interest when doing annotation in zoom mode
- Tutorials
A list of tutorials are available for users reference
V2 Alpha 2 Release, Renaming/Deleting data
Classifai V2.0.0-alpha2
12 August 2021
This version focuses more on bug fixing of the alpha1. Besides that, this version reenables some version 1 features to enhance annotation capability. This tools also starts supporting WebP image type.
Read alpha2 changelog for full information.
What's New:
- Rename data from workspace
The user can change the data file name from the workspace. All changes are reflected in the user folder.
- Delete data from workspace
The user can delete unwanted data from the workspace. Deleted data will not be removed completely from the dataset folder. Instead, it will be placed into a folder called .classifai_deleted_data. This will enable the user to restore the image if needed. Deleting this folder will remove the data completely from the dataset folder.
V2 Alpha Release, New User Interface and Functionalities
Classifai V2.0.0-alpha1
22 June 2021
What's New:
- New User Interface
Taking in user feedbacks, our teams have redesigned the user interface — which also gives us greater flexibility in integrating the new features you’ve been requesting such as:- New Project Tag
- Starred Project Tag
- Project Renaming
- Project Import Export
-
Project Import Export
-
Label Filtering
Now users can perform filtering of labels while exporting the output, this greatly reduces the amount of work for annotating the repeated images for different models.
What's Changed:
-
Project Creation Method
In ClassifAI v2, an image folder is required to be specified during project creation. This benefits user in multiple aspects, such as: -
Segmentation Temporarily Disabled
Image segmentation panel is still under development to comply with the new user interface. It will be enable again in the official release of V2.0.0. Stay tuned!
Improvements:
Shortcut key page, Malay language interface support, Chinese language interface enhancement, bug fix for #302
Classifai 1.2.0
26 February 2021
What's New:
-
Shorcut Key Page
Shortcut key page is now available, select the last button on the left panel to pop this window up
-
Malay language interface
Malay language interface is now supported by Classifai
Toggle to different languages in the home interface.
Enhancement:
- Chinese language interface enhancement
There were some keywords failed to covered in the mandarin version earlier.
Now, the chinese language is fully translated with an enhancement of the keywords to best explain the functionalities.
Before:
After:
Bug fixes:
- Load Project without Data (#302)
Previously progress update when loading project will occasionally complete without the data being fully loaded. This is now fixed.
Bug fix release for saving format, aspect ratio and more detailed logs
Classifai 1.1.1
29 January 2021
Critical Fixes:
-
Annotation output for bounding box projects changed to correspond to image name (#277)
The extension format of image, which previously concatenated into the annotation output of YOLO and Pascal VOC formats, is removed.Example
With image 1.jpg, the annotation output is 1.xml (Prior to fix: 1_jpg.xml)
With image sample.jpg, the annotation output is sample.txt (Prior to fix: sample_jpg.txt)
Other Fixes:
-
JPEG aspect ratio fix (#252)
Previously, some jpeg files contain EXIF orientation does not show the correct aspect ratio.
This is fixed in the current patch release. -
JPEG timeout fix (#280)
Loading images with large size for bounding box or segmentation projects has previously resulted in timeout error.
This problem has been solved. -
Enabling of detailed log messages if paths of images not found (#283)
When images previously imported could not be found in the system paths, the project will omit these images without causing any error. To give end users a better idea which images are missing, paths of images which cannot be found will be listed out in the log messages now.
v1.1.0: Delete feature for project and data points, bug fixes for #223
Classifai 1.1.0
11 January 2021
What's New:
- Deletion of project
- Deletion of data points by clicking on the images panel.
Note: Images selected will be removed from the project, but the images at the original path will stay intact.
- Segmentation Technique - Annotation + Shifting the Image
Bug Fixes:
Initial Release
Classifai 1.0.0
14 November 2020
What's New:
- Initial release of data annotation tool to public!
Major features:
-
Images Annotation with
- Bounding Box
- Classic Image Classification Problem:
- Create labels on the right panel and assign label to object of interest
- Text Recognition Use Cases (OCR):
- Text creation straight into the label box
- Classic Image Classification Problem:
- Polygons (Segmentation)
- Bounding Box
-
Conversion of files format for easy annotation
- PDF /TIF -> JPG/PNG
-
Distribution ready for easy installation on different operating systems:
- Windows (7, 8, 10)
- Mac
- Ubuntu (18 LTS, 20 LTS)
- Centos (7, 8)
Image Classification Labelling Method: Assign labels created from Label List
Text Recognition Labelling Method: Insert text into the blank label column