Image Segmentation

Draw polygon masks around objects for semantic segmentation tasks.

Configuration Fileconfig.yaml

annotation_task_name: "Image Segmentation"
task_name: "Image Segmentation"
task_description: "Draw precise polygon boundaries around objects."
task_dir: "."
port: 8000

data_files:
  - "sample-data.json"

item_properties:
  id_key: "id"
  text_key: "image_url"
  image_key: image_url

annotation_schemes:
  - annotation_type: image_annotation
    name: segments
    description: "Draw polygons around each object"
    tools:
      - polygon
    labels:
      - "Person"
      - "Vehicle"
      - "Animal"
      - "Object"

output_annotation_dir: "output/"
output_annotation_format: "json"

Sample Datasample-data.json

[
  {
    "id": "1",
    "image_url": "https://images.unsplash.com/photo-1518717758536-85ae29035b6d?w=640"
  },
  {
    "id": "2",
    "image_url": "https://images.unsplash.com/photo-1494976388531-d1058494cdd8?w=640"
  }
]

Get This Design

View on GitHub

Clone or download from the repository

Quick start:

git clone https://github.com/davidjurgens/potato-showcase.git
cd potato-showcase/templates/image/image-segmentation
potato start config.yaml

Details

Annotation Types

image_annotation

Domain

Computer Vision

Use Cases

SegmentationInstance Segmentation

Related Designs

Object Detection

Draw bounding boxes around objects for object detection model training.

image_annotation

RefCOCO - Referring Expression Grounding

Visual grounding task where annotators draw bounding boxes around objects referred to by natural language expressions. Based on the RefCOCO dataset (Yu et al., ECCV 2016), this task links referring expressions to spatial regions in images.

image_annotationtext

ViTPose Human Keypoint Annotation

Annotate human body keypoints and pose categories in images based on the ViTPose framework. Annotators mark anatomical landmarks, draw bounding boxes around people, and classify the overall pose type.

image_annotationradio