mb-s5mars

A segmentation dataset for planetary science applications.

Dataset Metadata

• License: CC-BY-4.0 (Creative Commons Attribution 4.0 International)
• Version: 1.0
• Date Published: 2025-05-15
• Cite As: TBD

Classes

This dataset contains the following classes:

• 0: Background
• 1: Bedrock
• 2: Hole
• 3: Ridge
• 4: Rock
• 5: Rover
• 6: Sand / Soil
• 7: Sky
• 8: Track

Directory Structure

The dataset follows this structure:

dataset/
  ├── train/
  │   ├── images/  # Image files
  │   └── masks/   # Segmentation masks
  ├── val/
  │   ├── images/  # Image files
  │   └── masks/   # Segmentation masks
  ├── test/
  │   ├── images/  # Image files
  │   └── masks/   # Segmentation masks

Statistics

• train: 4997 images
• val: 200 images
• test: 800 images
• partition_train_0.01x_partition: 49 images
• partition_train_0.02x_partition: 99 images
• partition_train_0.50x_partition: 2498 images
• partition_train_0.20x_partition: 999 images
• partition_train_0.05x_partition: 249 images
• partition_train_0.10x_partition: 499 images
• partition_train_0.25x_partition: 1249 images

Usage

from datasets import load_dataset

dataset = load_dataset("Mirali33/mb-s5mars")

Format

Each example in the dataset has the following format:

{
  'image': Image(...),      # PIL image
  'mask': Image(...),       # PIL image of the segmentation mask
  'width': int,             # Width of the image
  'height': int,            # Height of the image
  'class_labels': [str,...] # List of class names present in the mask
}