Back to datasets
Dataset assetOpen Source CommunityUnderwater Laser ImagingUCLR System

Coder-AN/StreakNet-Dataset

StreakNet‑Dataset is an underwater laser imaging dataset for the UCLR system. It contains stripe‑tube images captured by the UCLR system at distances of 10 m, 13 m, 15 m, and 20 m. Image resolution is 2048×2048 and the data type is uint16. The dataset is split into training, validation, and test sets: at 10 m there are 400 images, at 13 m 349 images, at 15 m 300 images, and at 20 m 267 images. The directory structure includes image data for each distance, ground‑truth files, preview images, and configuration files.

Source
hugging_face
Created
Nov 28, 2025
Updated
Apr 13, 2024
Signals
128 views
Availability
Linked source ready
Overview

Dataset description and usage context

StreakNet‑Dataset Overview

Dataset Description

StreakNet‑Dataset is a specialized underwater laser imaging dataset designed for the UCLR system. It comprises a series of stripe‑tube images captured at various distances (10 m, 13 m, 15 m, 20 m) using the UCLR system.

Detailed Information

DistanceNumber of Stripe‑tube ImagesResolutionData TypeTraining SetValidation SetTest Set
10 m4002048×2048uint16315,20040,800819,200
13 m3492048×2048uint16281,99247,530714,752
15 m3002048×2048uint16245,40039,200614,400
20 m2672048×2048uint16229,08631,240546,816

Dataset Download

StreakNet‑Dataset can be downloaded for free from HuggingFace or ModelScope.

Dataset Organization

After downloading, the directory layout of StreakNet‑Dataset is as follows:

sh
datasets
    |- clean_water_10m      # Data directory for 10 m distance
    |   |- data             # Raw stripe images
    |   |   |- 001.tif
    |   |   |- 002.tif
    |   |   |- 003.tif
    |   |   |- ...
    |   |
    |   |- groundtruth.npy  # Ground‑truth image
    |   |- preview.jpg      # Preview of ground‑truth
    |
    |- clean_water_13m      # 13 m distance (same structure as 10 m)
    |- clean_water_15m      # 15 m distance (same structure as 10 m)
    |- clean_water_20m      # 20 m distance (same structure as 10 m)
    |- template.npy         # 1‑D time‑series of the template signal
    |- test_config.yaml     # Test set configuration file
    |- train_config.yaml    # Training set configuration file
    |- valid_config.yaml    # Validation set configuration file
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