Releases: IDEA-Research/deepdataspace
Releases · IDEA-Research/deepdataspace
v0.9.0
What's New:
- refactor: switch from JSON to Python script for COCO meta file.
- feature: enable caption visualization for COCO datasets.
- feature: allow custom function for caption generation in COCO meta file.
- bugfix: fix object saving issue for last image in saving batch.
本次更新:
- 重构: 将 COCO meta 文件从 JSON 切换为 Python 脚本。
- 功能: 支持 COCO 数据集的 caption 可视化。
- 功能: 在 COCO meta 文件中允许自定义 caption 生成函数。
- 修复: 解决批量保存图片时,最后一张图片对象未正确保存的问题。
v0.8.0
What's New
- feat: remove the restriction on the number of annotations saved in collaborative annotation
- fix: fix dataset tiled compare list styles
- fix: fix user lint api to user name
- fix: fix missing annotations when importing a dataset
- fix: fix the api error when saving annotations in collaborative annotation
本次更新
- 功能:去除协同标注的标注数目保存限制
- 修复:修复数据集平铺对比的UI显示问题
- 修复:修复协同标注用户搜索逻辑
- 修复:修复导入数据集时缺失部分 annotation 的问题
- 修复:修复协同标注中保存标注时的接口异常
v0.7.0
What's New
- refactor: refactor frontend project structure
- refactor: support import coco2017 dataset by a json config file
- fix: fix annotation bug in label editor
- fix: fix fnfp match logic in dataset viewer
- fix: fix error of reading 'score' value from 'conf' field for coco2017 importer
- chore: fix build files for frontend project
本次更新
- 重构:重构前端项目架构
- 重构:导入 coco2017 数据集时支持通过 json 文件指定各文件
- 修复:修复标注编辑器中的一些问题
- 修复:修复数据集可视化中的 fnfp 匹配问题
- 修复:修复导入 coco2017 数据集时错误地从 conf 字段读取 score 值的问题
- 构建:修复前端构建文件
v0.6.0
What's New
- feat: add support for the mask annotation type in the quick annotation mode
- fix: fix the logic for limiting the status of collaborative annotation tasks
- optimize: refactor the code of the annotation editor and optimize the related user experience
本次更新
- 功能:快速标注模式新增支持 mask标注类型
- 修复:修复协同标注任务状态限制逻辑
- 优化:重构标注编辑器实现及优化相关体验
v0.5.0
What's New
- feat: add detailed error codes and error messages on all scenarios
- feat: support for directly trying out the intelligent annotation APIs from the DDS algorithm platform
- feat: add prompt message for quick mode annotation
- optimize: download sample datasets from accelerated host
- fix: upgrade open-source components and dependencies to fix security vulnerabilities for frontend and backend
- chore: upload test report to codecov in tool-ci pipeline
- docs: refine API docs for all models definition
本次更新
- 功能:新增各场景下的错误码和错误消息提示
- 功能:支持直接体验 DDS 算法平台的智能标注
- 功能:增加快速标注提示文案
- 优化:优化快速启动时下载样例数据集的速度
- 修复:升级前后端开源组件及依赖以修复安全漏洞
- 构建:恢复上传单元测试的报告到 codecov
- 文档:优化各 Model 定义的 API 文档
v0.4.1
v0.4.0
What's New
- feat: support launch DDS with Docker Compose
- feat: add description for dataset list
- optimize: render entire workspace with Canvas in annotation editor
- fix: display issue of dataset card with null cover image
本次更新
- 功能: 支持使用 Docker Compose 启动 DDS
- 功能: 为数据集列表添加描述
- 优化: 在标注编辑器中使用 Canvas 渲染整个工作区
- 修复: 数据集封面卡片显示问题
v0.3.0
What's New
- feat: support displaying dataset covers
- feat: support exporting label project back to datasets
- feat: support specifying a string or a function as dataset description
- feat: add kindly reminder for mobile display
- fix: fix compatibility problem for celery worker on Win10 platform
- optimize: Improve API error handler logic and error code message
- docs: highlight platform compatibility in project README
本次更新
- 功能: 支持展示数据集封面
- 功能: 支持导出协同标注的结果到数据集中
- 功能: 支持定义数据集的描述信息
- 功能: 增加手机端访问提示
- 修复: 修复 Win10 平台兼容性问题
- 优化: 优化接口异常处理逻辑及错误码提示信息
- 文档: 项目README中高亮平台兼容性内容
v0.2.0
What's New
- feat: Support importing pre-labels for label project
- feat: Support polling AI annotate APIs(detect, segment, pose)
- feat: Support running DDS in Docker
- bugfix: Fix some compatibility problems on Windows and Mac
- test: Add frontend unit tests
- test: Include new platforms for Python test (Windows, Mac)
- chore: Optimize CI and CD pipelines
本次更新
- 功能:支持为标注项目导入预标注数据
- 功能:支持轮询 AI 标注 API(检测、分割、姿态)
- 功能:支持在 Docker 中运行 DDS
- 修复:修复 Windows 和 Mac 上的一些兼容性问题
- 测试:添加前端单元测试
- 测试:为 Python 测试包含新平台(Windows、Mac)
- 构建:优化 CI 和 CD 流水线