Back to datasets
Dataset assetOpen Source CommunityComputer VisionImage Stitching

UDIS-D

The UDIS‑D dataset provides images and masks for unsupervised deep image stitching. It is split into training and testing sets, each containing multiple images and their corresponding masks.

Source
github
Created
Nov 5, 2024
Updated
Nov 28, 2024
Signals
807 views
Availability
Linked source ready
Overview

Dataset description and usage context

Dataset Overview

Source

  • UDIS‑D Dataset: Available from UDIS with aligned images and masks.

Structure

The dataset should be organized as follows:

Training Set

train
├── input1
│   ├── 000001.jpg
│   └── ...
├── input2
│   ├── 000001.jpg
│   └── ...
├── mask1
│   ├── 000001.jpg
│   └── ...
├── mask2
│   ├── 000001.jpg
│   └── ...

Test Set

test
├── warp1
│   ├── 000001.jpg
│   └── ...
├── warp2
│   ├── 000001.jpg
│   └── ...
├── mask1
│   ├── 000001.jpg
│   └── ...
├── mask2
│   ├── 000001.jpg
│   └── ...
Need downstream help?

Pair the dataset with AI analysis and content workflows.

Once the source passes your review, move straight into summarization, transformation, report drafting, or presentation generation with the JuheAI toolchain.

Explore AI studio