This is a joint work with the Ohio State University and Meta Reality Labs Research.</p>\n<p>We introduce SuperMemory-VQA, an egocentric visual question answering (VQA) dataset for evaluating AI assistants on practical, long-horizon memory tasks. </p>\n<ul>\n<li>It contains 52.9 hours of everyday activities recorded with Meta's Project Aria glasses, including synchronized RGB video, audio transcription, eye gaze, IMU, and SLAM trajectories. </li>\n<li>Through a human-verified annotation pipeline, we construct grounded 4,853 question-answer pairs that span object and location memory, intent recall, visual scene recall, timeline reconstruction, conversational memory, and in-context retrieval. </li>\n<li>Our participant survey further supports that our questions are realistic, useful, and aligned with everyday memory needs.</li>\n</ul>\n","updatedAt":"2026-06-04T18:11:31.281Z","author":{"_id":"6526e45d2d07ec2439d4f112","avatarUrl":"/avatars/e86b29f35acf5311681445917d59b0f9.svg","fullname":"Hyo Jin Kim","name":"hyojinie","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":1,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.8305431604385376},"editors":["hyojinie"],"editorAvatarUrls":["/avatars/e86b29f35acf5311681445917d59b0f9.svg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2606.00825","authors":[{"_id":"6a20ab6d15100c5272a84583","name":"Samiul Alam","hidden":false},{"_id":"6a20ab6d15100c5272a84584","name":"Shakhrul Iman Siam","hidden":false},{"_id":"6a20ab6d15100c5272a84585","name":"Michael J. Proulx","hidden":false},{"_id":"6a20ab6d15100c5272a84586","name":"James Fort","hidden":false},{"_id":"6a20ab6d15100c5272a84587","name":"Richard Newcombe","hidden":false},{"_id":"6a20ab6d15100c5272a84588","user":{"_id":"6526e45d2d07ec2439d4f112","avatarUrl":"/avatars/e86b29f35acf5311681445917d59b0f9.svg","isPro":false,"fullname":"Hyo Jin Kim","user":"hyojinie","type":"user","name":"hyojinie"},"name":"Hyo Jin Kim","status":"claimed_verified","statusLastChangedAt":"2026-06-04T12:40:39.955Z","hidden":false},{"_id":"6a20ab6d15100c5272a84589","name":"Mi Zhang","hidden":false}],"publishedAt":"2026-05-30T00:00:00.000Z","submittedOnDailyAt":"2026-06-04T00:00:00.000Z","title":"SuperMemory-VQA: An Egocentric Visual Question-Answering Benchmark for Long-Horizon Memory","submittedOnDailyBy":{"_id":"6526e45d2d07ec2439d4f112","avatarUrl":"/avatars/e86b29f35acf5311681445917d59b0f9.svg","isPro":false,"fullname":"Hyo Jin Kim","user":"hyojinie","type":"user","name":"hyojinie"},"summary":"AI glasses present a compelling platform for AI agents to serve as personalized memory assistants. To be genuinely useful, such systems must move beyond short-term video comprehension and address memory gaps that humans experience for practical, personal, or social purposes over longitudinal egocentric video streams. However, existing egocentric datasets predominantly focus on action recognition or generic QAs from short clips, measuring perceptual capabilities rather than realistic human memory needs. We introduce SuperMemory-VQA, an egocentric visual question answering (VQA) dataset for evaluating AI assistants on practical, long-horizon memory tasks. It contains 52.9 hours of everyday activities recorded with AI glasses, including synchronized RGB video, audio transcription, eye gaze, IMU, and SLAM trajectories. Through a human-verified annotation pipeline, we construct grounded 4,853 question-answer pairs that span object and location memory, intent recall, visual scene recall, timeline reconstruction, conversational memory, and in-context retrieval. Each question is posed as multiple-choice with an explicit \"unanswerable\" option to test hallucination robustness. Benchmarking leading agentic frameworks and LLM backbones reveals that existing systems remain far from reliable on real-world memory tasks, highlighting the need for new architectures for grounded AI memory that can answer only when evidence is sufficient. A participant survey further supports that our questions are realistic, useful, and aligned with everyday memory needs.","upvotes":1,"discussionId":"6a20ab6d15100c5272a8458a","projectPage":"https://supermemory-vqa.github.io/","githubRepo":"https://github.com/AIoT-MLSys-Lab/supermemory-vqa","githubRepoAddedBy":"user","ai_summary":"SuperMemory-VQA is introduced as an egocentric visual question answering dataset designed to evaluate AI assistants on long-term memory tasks through real-world activities recorded with AI glasses.","ai_keywords":["egocentric visual question answering","AI glasses","personalized memory assistants","longitudinal egocentric video streams","VQA dataset","RGB video","audio transcription","eye gaze","IMU","SLAM trajectories","grounded question-answer pairs","object memory","location memory","intent recall","visual scene recall","timeline reconstruction","conversational memory","in-context retrieval","hallucination robustness","agentic frameworks","LLM backbones","real-world memory tasks"],"ai_summary_model":"Qwen/Qwen2.5-Coder-32B-Instruct","githubStars":1,"organization":{"_id":"64f00e771205f43449300336","name":"ohiostate","fullname":"The Ohio State University","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/64f00dab77777df14e39d0b4/suMjJQTVx2mnCW9u3vahz.png"}},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"6526e45d2d07ec2439d4f112","avatarUrl":"/avatars/e86b29f35acf5311681445917d59b0f9.svg","isPro":false,"fullname":"Hyo Jin Kim","user":"hyojinie","type":"user"}],"acceptLanguages":["en"],"dailyPaperRank":0,"organization":{"_id":"64f00e771205f43449300336","name":"ohiostate","fullname":"The Ohio State University","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/64f00dab77777df14e39d0b4/suMjJQTVx2mnCW9u3vahz.png"},"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2606/2606.00825.md"}">
SuperMemory-VQA: An Egocentric Visual Question-Answering Benchmark for Long-Horizon Memory
Abstract
SuperMemory-VQA is introduced as an egocentric visual question answering dataset designed to evaluate AI assistants on long-term memory tasks through real-world activities recorded with AI glasses.
AI glasses present a compelling platform for AI agents to serve as personalized memory assistants. To be genuinely useful, such systems must move beyond short-term video comprehension and address memory gaps that humans experience for practical, personal, or social purposes over longitudinal egocentric video streams. However, existing egocentric datasets predominantly focus on action recognition or generic QAs from short clips, measuring perceptual capabilities rather than realistic human memory needs. We introduce SuperMemory-VQA, an egocentric visual question answering (VQA) dataset for evaluating AI assistants on practical, long-horizon memory tasks. It contains 52.9 hours of everyday activities recorded with AI glasses, including synchronized RGB video, audio transcription, eye gaze, IMU, and SLAM trajectories. Through a human-verified annotation pipeline, we construct grounded 4,853 question-answer pairs that span object and location memory, intent recall, visual scene recall, timeline reconstruction, conversational memory, and in-context retrieval. Each question is posed as multiple-choice with an explicit "unanswerable" option to test hallucination robustness. Benchmarking leading agentic frameworks and LLM backbones reveals that existing systems remain far from reliable on real-world memory tasks, highlighting the need for new architectures for grounded AI memory that can answer only when evidence is sufficient. A participant survey further supports that our questions are realistic, useful, and aligned with everyday memory needs.
Community
This is a joint work with the Ohio State University and Meta Reality Labs Research.
We introduce SuperMemory-VQA, an egocentric visual question answering (VQA) dataset for evaluating AI assistants on practical, long-horizon memory tasks.
- It contains 52.9 hours of everyday activities recorded with Meta's Project Aria glasses, including synchronized RGB video, audio transcription, eye gaze, IMU, and SLAM trajectories.
- Through a human-verified annotation pipeline, we construct grounded 4,853 question-answer pairs that span object and location memory, intent recall, visual scene recall, timeline reconstruction, conversational memory, and in-context retrieval.
- Our participant survey further supports that our questions are realistic, useful, and aligned with everyday memory needs.
Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.
Tap or paste here to upload images
Cite arxiv.org/abs/2606.00825 in a model README.md to link it from this page.
Cite arxiv.org/abs/2606.00825 in a dataset README.md to link it from this page.
Cite arxiv.org/abs/2606.00825 in a Space README.md to link it from this page.
Discussion (0)
Sign in to join the discussion. Free account, 30 seconds — email code or GitHub.
Sign in →No comments yet. Sign in and be the first to say something.