Weather-nuScenes Dataset Grid Viewer

This project is a static image viewer for the dataset_test folder with a sample of the Weather-nuScenes dataset that can be accessed in: https://doi.org/10.5281/zenodo.21135950

It shows:

• weather conditions on the X axis
• cameras on the Y axis
• one comparison grid per photo group
• pagination with 5 photo groups per page

How to open the viewer

The viewer is fully static.

1. Open index.html with a double click.
2. The page loads manifest.js directly.
3. Images are resolved from the dataset_test folder using relative paths.

No server is required.

Project files

• index.html: page structure
• styles.css: layout and visual styling
• script.js: grid rendering, pagination, hover actions, and comparison modal
• manifest.js: static data used by the viewer
• build_manifest.js: script that updates manifest.js

How manifest.js works

The manifest is built from the image order in dataset_test/0_originals.

It stores:

• photos: array of photo groups
• group: the unique row id for that image index
• originals: the original filename for each camera

The viewer builds the rest of the file paths in the browser from those original names.

Updating the manifest

If you add new files to dataset_test, run:

node build_manifest.js

What the script does:

• scans dataset_test/0_originals
• sorts the images in each camera folder
• uses the image position to create one group per row
• updates manifest.js
• writes one manifest entry per image index

Expected dataset structure

The viewer expects this structure:

dataset_test/
  0_originals/
    CAM_BACK/
    CAM_BACK_LEFT/
    CAM_BACK_RIGHT/
    CAM_FRONT/
    CAM_FRONT_LEFT/
    CAM_FRONT_RIGHT/
  fog/
  night/
  night_rain/
  night_snow/
  rain/
  sandstorm/
  snow/

The browser code derives variant filenames from the original image names using the current naming pattern.