README.md

metadata

license: other
task_categories:
  - image-segmentation
tags:
  - cvat
  - segmentation
size_categories:
  - 1K<n<10K

Sinon_Less - Segmentation Dataset

Instance Segmentation dataset with polygon annotations.

Dataset Description

• Source: CVAT Annotation Platform
• Format: Segmentation
• Total Samples: 2535
• Splits: train: 1497, val: 267, test: 771

Labels

• Person
• Hard Hat
• No Head Protection
• Cleanroom Suit
• Splash Proof Gown
• No Protective Clothing
• Full Face Mask
• No Full Face Mask

Usage

from datasets import load_dataset

# Load the dataset
dataset = load_dataset("Daniel246/cvat-sinon_less-segmentation")

# Access splits
train_data = dataset["train"]
val_data = dataset["val"]
test_data = dataset["test"]

# Example: iterate over samples
for sample in train_data:
    image = sample["image"]
    # Process annotations based on format type

Dataset Structure

Data Fields

• image: PIL Image
• image_id: Unique identifier for the image
• width: Image width in pixels
• height: Image height in pixels
• objects: List of segmented objects with:
  • bbox: Bounding box [x_center, y_center, width, height] (normalized)
  • polygon: List of [x, y] points (normalized)
  • category_id: Class index
  • category_name: Class name

License

Please check the original data source for licensing information.

---

Generated by CVAT Training System