Release v0.9.1

A few bug fixes and optimizations.

1. [fix #216] 2D auxiliary lines and point cloud RGB support.
2. [fix #223] Request URI too large.
3. [fix #240] TempDataId duplicated when uploading dataset.
4. [fix #234] AnnotationStatus of data not change when submit.
5. Increase nginx body size limit to 1TB.
6. Do not load all content into browser memory when uploading file.

Release v0.9

1. A completely rewritten image annotation tool；
2. Support continuous frame annotation of images；
3. A few small enhancements;

Release v0.8.1

1. Fix scene upload error.
2. Customize mysql configuration to allow 1GB package.

Release v0.8

1. Point cloud data frames (scene) annotation, uploading data frames to the dataset through zip package and annotating them in the tool;
2. Adjust the directory structure of the dataset zip package to be consistent with BasicAI Cloud;
3. A few bug fixes;

Release v0.7.3

1. Specify image pull policy of all model service to Always in docker compose to avoid image caching.
2. A few docs related update.

Release v0.7.2

1. Fixed the issue of missing Class in image object detection model;
2. Fixed the abnormal login status issue of the point cloud tool;
3. Optimized document.

Release v0.7.1

Fix bugs

1. Fixed the problem of exporting no class after importing class;
2. Fixed the problem that the MinIO pre-signed address check fails;
3. Fixed the problem that Chinese characters appear in the url;
4. Fixed data result and add classID;
5. Fixed text-dataset load bug;
6. Fixed upload by url.

Release v0.7.0

Release v0.7.0

What's new?

Multiple rounds of dialogues annotation refer to the process of annotating a conversation between two or more parties through multiple iterations or rounds of annotation. In this process, after annotating each round of dialogue, the annotations are reviewed and corrected if necessary before proceeding to the next round of annotation. This iterative process is often used when handling complex or nuanced conversations that require careful analysis and interpretation. Multiple rounds of dialogue annotation can help ensure data quality and accuracy by allowing annotators to review their work and make corrections before moving on to subsequent rounds. It also helps to improve inter-annotator agreement and reduces errors and inconsistencies in the final annotated dataset.

Multiple rounds of dialogues annotation

• Add new data category - Text is support
• Allow user to upload text datasets
• New thumbnails and UI styles

This is still an early version of text annotation for LLM.
Sample dataset can be download at here.

Release v0.6.1

In v0.6.1, many bugs are fixed:

1. Fixed import and export bugs of coco format and results;
2. Fixed the data duplication bug in the data list;
3. Fixed the bug that the id of the exported class was missing;
4. Click the document button to jump to a new page;
5. Fixed the error, when save point cloud results.

Release v0.6.0

Release v0.6.0

New Features

1. Model Integration

To give users more control over their own models, we have added functionality that allows them to create, edit, search, and delete models. With this feature, users can manage basic model information, classes, configuration details, and run the model on datasets.
For more information, please refer to our Model Integration Guide.
https://docs.xtreme1.io/xtreme1-docs/developer-reference/model-integration-guide

2. Enhanced the Function of Data Import and Export

Data Import:

• Support for importing pre-annotation dataset on 3D Dataset
  Note: Only importing the class name is supported, attributes and global classification are not yet supported.

Data Export:

• COCO format can be exported directly
• When exporting data, the filter can be applied.

3. Image Similarity Map Optimization

• The algorithm is optimized.
• Selecting multiple targets in the image similarity map, you can jump to the data tab to find the corresponding images.

4. Add Dataset Split

In the data preparation process of model training, the data can be directly divided into sub-datasets of Training, Test, and Validation sets.

Changes and Bug Fixes

• Optimized the clarity of the preview image, including in dataset and scenario search.
• In Scenario Search, the search box style is changed.
• Fixed the result could not be calculated in some cases.

As always, if you have any feedback or suggestions for improvement, please don't hesitate to let us know.