图像数据标注工具推荐-CVAT

摘要： 从事深度学习研究开始，尝试过很多不同类型，不同功能的数据标注工具，尤其是图像标注工具，包括自己写的简单工具，都不能很好的满足任务的需求，近期突然使用了CVAT，感觉似乎可能成为未来工具包中的常备系列，写此一文，记录使用过程。

Computer Vision Annotation Tool (CVAT) is a web-based tool which helps to annotate videos and images for Computer Vision algorithms. It was inspired by Vatic free, online, interactive video annotation tool. CVAT has many powerful features: interpolation of bounding boxes between key frames, automatic annotation using deep learning models, shortcuts for most of critical actions, dashboard with a list of annotation tasks, LDAP and basic authorization, etc… It was created for and used by a professional data annotation team. UX and UI were optimized especially for computer vision tasks developed by our team.

安装部署与配置

部署过程参考官方文档：CVAT 部署文档，写的极为详细，基本没有太多坑🕳。

由于本身的需求，需要使用一些特定功能，具体如下：

1. 配置外网访问

构建YAML文件：docker-compose.add-on.yml:

1
2
3
4
5
6

version: "2.3"

services:
  cvat_proxy:
    environment:
      CVAT_HOST: .example.com

2. 配置挂载图像存储目录

替换{}中目录位置

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15

version: "2.3"

services:
  cvat:
    environment:
      CVAT_SHARE_URL: "Mounted from /mnt/share host directory"
    volumes:
      - cvat_share:/home/django/share:ro

volumes:
  cvat_share:
    driver_opts:
      type: none
      device: {/mnt/share}
      o: bind

完成上述配置，重新执行build过程：

1

docker-compose -f docker-compose.yml -f docker-compose.add-on.yml up -d --build

3. 开启CUDA/OpenVINO support | 分析组件

1
2
3
4
5
6

# Build and run containers with CUDA and OpenVINO support
# IMPORTANT: need to download OpenVINO package before running the command
docker-compose -f docker-compose.yml -f components/cuda/docker-compose.cuda.yml -f components/openvino/docker-compose.openvino.yml up -d --build

# Build and run containers with Analytics component support:
docker-compose -f docker-compose.yml -f components/analytics/docker-compose.analytics.yml up -d --build

其它命令

1
2

# stops and removes containers, networks, volumes, and images
docker-compose down

使用技巧

相关地址

任务

每个数据集的标注工作都作为一项任务；

支持两种工作模式：

• interpolation task (video sequence)
• annotation task (independent images)

快捷键

常用快捷键：

• N：快速创建BOX标记工具；
• Crtl + S: 随时保存
• D： 上一张
• F： 下一张
• CRTL + #N（标签序号）

Hints:

Hold Mouse Wheel to move frame (for example, while drawing)

Hold Ctrl when shape is active and fix it.

Hold Ctrl when paste shape from buffer for multiple pasting.

半监督标记与模型

To-Do

• Auto annotation using DL models in OpenVINO toolkit format
• Analytics: management and monitoring of data annotation team
• TF Object Detection API: auto annotation
• Support for NVIDIA GPUs
• Semi-automatic segmentation with Deep Extreme Cut
• Auto segmentation: Keras+Tensorflow Mask R-CNN Segmentation

参考

1. CVAT Github
2. CVAT 部署文档
3. CVAT 用户文档