ImageEval 2026 — Ayn-VQA Task 1b: Hallucination Detection (English)
LiveAI & MLAR / VR

ImageEval 2026 — Ayn-VQA Task 1b: Hallucination Detection (English)

Given an image and 3 statements (English), label EACH statement True (grounded in the image) or False (hallucinated). Exactly one statement is grounded. Part of the ImageEval 2026 ...

hunzedOrganizer hunzed
Official site

About this hackathon

Given an image and 3 statements (English), label EACH statement True (grounded in the image) or False (hallucinated). Exactly one statement is grounded. Part of the ImageEval 2026 Ayn-VQA shared task on cultural grounding in Arabic multimodal understanding. Data on HuggingFace (QCRI/AynVQA-ArabicNLP26). Join our Slack for announcements and discussion: https://join.slack.com/t/mm-eval/shared_invite/zt-41j09ml4j-WMn0NzhqAgT9ZK6e1L8eBA

Tracks

General Track

Given an image and 3 statements (English), label EACH statement True (grounded in the image) or False (hallucinated). Exactly one statement is grounded. Part of the ImageEval 2026 Ayn-VQA shared task on cultural grounding in Arabic multimodal understanding. Data on HuggingFace (QCRI/AynVQA-ArabicNLP26). Join our Slack for announcements and discussion: https://join.slack.com/t/mm-eval/shared_invite/zt-41j09ml4j-WMn0NzhqAgT9ZK6e1L8eBA

Prizes

1

Project Prize

Given an image and 3 statements (English), label EACH statement True (grounded in the image) or False (hallucinated). Exactly one statement is grounded. Part of the ImageEval 2026 Ayn-VQA shared task on cultural grounding in Arabic multimodal understanding. Data on HuggingFace (QCRI/AynVQA-ArabicNLP26). Join our Slack for announcements and discussion: https://join.slack.com/t/mm-eval/shared_invite/zt-41j09ml4j-WMn0NzhqAgT9ZK6e1L8eBA

$1,000

Schedule

  1. May 21, 04:00 PM

Tags

#Codabench#AI#Competition#competition