Loading...

Caption Scorer

Detect misleading media, verify article context, and flag clickbait in seconds.

How the Caption Scorer works

Caption Scorer checks how well a photo's caption actually matches what's shown in the image, using a match score and a visual heatmap you can point to as evidence.

  1. STEP 01

    Pick a starting point

    Use Manual Upload to score your own image and caption, or Analyse via Article URL to pull the hero image, title, and caption straight from a published article.

  2. STEP 02

    Add the caption details

    For a manual upload, add the image plus a caption and an optional headline. For a URL, just paste the link, no extra typing needed.

  3. STEP 03

    Run the analysis

    The tool scores how closely the caption matches the image, and does the same for the headline if one was provided.

  4. STEP 04

    Read the match score

    Each score reflects how well the caption or headline lines up with what's actually in the photo. The higher the score, the stronger the match.

  5. STEP 05

    Check the heatmap

    The highlighted overlay shows exactly which part of the image the caption or headline draws attention to, so you can see where it lines up or falls short.

Pro Tip

Getting the clearest read

If both a caption and headline are scored, compare their heatmaps side by side. A mismatch between the two often points to where the story and the image diverge.

Tips for Best Results

Be specific: Include sport, timeframe, or location where possible

Ask about trends, representation, tone, or comparisons

Use plain language — no special syntax required

ICT AI Logo
Partner Logo
MDIA Logo
Threls Logo

Connect

Follow

© 2026 University of Malta. All rights reserved.