Live
Black Hat USADark ReadingBlack Hat AsiaAI BusinessCollege students say they are changing their majors because of AIBusiness InsiderInside KPMG's push to turn tax experts into hands-on software buildersBusiness Insider'You’re not trying to be perfect': I tested ChatGPT’s advice for burnout — and it helped - TechRadarGoogle News: ChatGPTU.S. Postal Inspection Service warns public about new wave of scams powered by artificial intelligence - WMAR 2 News BaltimoreGNews AI USAGame to Lose Online Mode After Its Server Partner Pivots to You’ll Never Guess WhatGizmodoThis $400 (Not) AI Keychain Is Pointless, Extravagant, and Weirdly LovableGizmodoI used ChatGPT to transform my iPhone home screen — and now it feels like a brand-new device - Tom's GuideGoogle News: ChatGPTUganda to Unveil Comprehensive AI and Emerging Tech Roadmap by June - PC Tech MagazineGoogle News - AI UgandaHow Israel is expanding its use of AI warfare in Iran and Lebanon - Al JazeeraGNews AI IsraelPresentation: Directing a Swarm of Agents for Fun and ProfitInfoQ AI/MLForthcoming machine learning and AI seminars: April 2026 editionAIhubAssembly of 59 Best of Sensors 2026 finalists announcedFierce ElectronicsBlack Hat USADark ReadingBlack Hat AsiaAI BusinessCollege students say they are changing their majors because of AIBusiness InsiderInside KPMG's push to turn tax experts into hands-on software buildersBusiness Insider'You’re not trying to be perfect': I tested ChatGPT’s advice for burnout — and it helped - TechRadarGoogle News: ChatGPTU.S. Postal Inspection Service warns public about new wave of scams powered by artificial intelligence - WMAR 2 News BaltimoreGNews AI USAGame to Lose Online Mode After Its Server Partner Pivots to You’ll Never Guess WhatGizmodoThis $400 (Not) AI Keychain Is Pointless, Extravagant, and Weirdly LovableGizmodoI used ChatGPT to transform my iPhone home screen — and now it feels like a brand-new device - Tom's GuideGoogle News: ChatGPTUganda to Unveil Comprehensive AI and Emerging Tech Roadmap by June - PC Tech MagazineGoogle News - AI UgandaHow Israel is expanding its use of AI warfare in Iran and Lebanon - Al JazeeraGNews AI IsraelPresentation: Directing a Swarm of Agents for Fun and ProfitInfoQ AI/MLForthcoming machine learning and AI seminars: April 2026 editionAIhubAssembly of 59 Best of Sensors 2026 finalists announcedFierce Electronics
AI NEWS HUBbyEIGENVECTOREigenvector

Storing Less, Finding More: How Novelty Filtering Improves Cross-Modal Retrieval on Edge Cameras

arXiv cs.IRby Sherif AbdelwahabApril 1, 20261 min read0 views
Source Quiz

arXiv:2603.29631v1 Announce Type: cross Abstract: Always-on edge cameras generate continuous video streams where redundant frames degrade cross-modal retrieval by crowding correct results out of top-k search. This paper presents a streaming retrieval architecture: an on-device epsilon-net filter retains only semantically novel frames, building a denoised embedding index; a cross-modal adapter and cloud re-ranker compensate for the compact encoder's weak alignment. A single-pass streaming filter outperforms offline alternatives (k-means, farthest-point, uniform, random) across eight vision-language models (8M-632M) on two egocentric datasets (AEA, EPIC-KITCHENS). Combined, the architecture reaches 45.6% Hit@5 on held-out data using an 8M on-device encoder at an estimated 2.7 mW.

View PDF HTML (experimental)

Abstract:Always-on edge cameras generate continuous video streams where redundant frames degrade cross-modal retrieval by crowding correct results out of top-k search. This paper presents a streaming retrieval architecture: an on-device epsilon-net filter retains only semantically novel frames, building a denoised embedding index; a cross-modal adapter and cloud re-ranker compensate for the compact encoder's weak alignment. A single-pass streaming filter outperforms offline alternatives (k-means, farthest-point, uniform, random) across eight vision-language models (8M-632M) on two egocentric datasets (AEA, EPIC-KITCHENS). Combined, the architecture reaches 45.6% Hit@5 on held-out data using an 8M on-device encoder at an estimated 2.7 mW.

Comments: 6 pages, 3 figures, 5 tables; supplementary video included as ancillary file

Subjects:

Computer Vision and Pattern Recognition (cs.CV); Distributed, Parallel, and Cluster Computing (cs.DC); Information Retrieval (cs.IR)

ACM classes: I.4.9; I.2.10

Cite as: arXiv:2603.29631 [cs.CV]

(or arXiv:2603.29631v1 [cs.CV] for this version)

https://doi.org/10.48550/arXiv.2603.29631

arXiv-issued DOI via DataCite (pending registration)

Submission history

From: Sherif Abdelwahab [view email] [v1] Tue, 31 Mar 2026 11:54:38 UTC (8,495 KB)

Was this article helpful?

Sign in to highlight and annotate this article

AI
Ask AI about this article
Powered by Eigenvector · full article context loaded
Ready

Conversation starters

Ask anything about this article…

Daily AI Digest

Get the top 5 AI stories delivered to your inbox every morning.

Knowledge Map

Knowledge Map
TopicsEntitiesSource
Storing Les…modellanguage mo…announcealignmentpaperarxivarXiv cs.IR

Connected Articles — Knowledge Graph

This article is connected to other articles through shared AI topics and tags.

Knowledge Graph100 articles · 143 connections
Scroll to zoom · drag to pan · click to open

Discussion

Sign in to join the discussion

No comments yet — be the first to share your thoughts!