frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

Open in hackernews

Show HN: Gemini can now natively embed video, so I built sub-second video search

https://github.com/ssrajadh/sentrysearch
60•sohamrj•2h ago
Gemini Embedding 2 can project raw video directly into a 768-dimensional vector space alongside text. No transcription, no frame captioning, no intermediate text. A query like "green car cutting me off" is directly comparable to a 30-second video clip at the vector level.

I used this to build a CLI that indexes hours of footage into ChromaDB, then searches it with natural language and auto-trims the matching clip. Demo video on the GitHub README. Indexing costs ~$2.50/hr of footage. Still-frame detection skips idle chunks, so security camera / sentry mode footage is much cheaper.

Comments

ygouzerh•1h ago
That's quite interesting, well done! I haven't thought of this use case for embeddings. It open the door to quite many potential applications!
stavros•1h ago
Man, the surveillance applications for this are staggering.
dev_tools_lab•1h ago
Nice use of native video embedding. How do you handle cases where Gemini's response confidence is low? Do you have a fallback or threshold?
sohamrj•1h ago
as of now, no threshold but that is planned in the future.

for example, for now if i search "cybertruck" in my indexed dashcam footage, i don't have any cybertrucks in my footage, so it'll return a clip of the next best match which is a big truck, but not a cybertruck

mdrzn•1h ago
Very interesting (not for a dashcam, but for home monitoring).
klntsky•1h ago
why not skip the text conversion? is it usable at all?
sohamrj•1h ago
gemini embedding 2 converts straight video to vectors. in this case, dashcam clips don't have audio to transcribe and even if they did, it would be useless in the search
password4321•46m ago
What are the SoA audio models right now?
Aeroi•1h ago
very cool, anybody have apparent use cases for this?
sohamrj•55m ago
dashcam and home security footage are the 2 main ones i can think of.

a bit expensive right now so it's not as practical at scale. but once the embedding model comes out of public preview, and we hopefully get a local equivalent, this will be a lot more practical.

hebelehubele•43m ago
State surveillance
wahnfrieden•16m ago
Worker surveillance
giozaarour•27m ago
I think a good use case would be searching for certain products or videos across social media (TikTok and Instagram). especially useful for shopping, maybe
vidarh•15m ago
Branding/marketing monitoring companies would be all over this.
emsign•47m ago
Where is the Exit to this dystopia?
BrokenCogs•30m ago
The Matrix style human pods: we live in blissful ignorance in the Matrix, while the LLMs extract more and more compute power from us so some CEO somewhere can claim they have now replaced all humans with machines in their business.
throwup238•17m ago
I was thinking more of the season 3 episode of Doctor Who titled Gridlock where everyone lives in flying cars circling a giant expressway underground, while all the upper class people on the surface died years ago from a pandemic.
ting0•6m ago
Ever get the feeling that the universe is reading your mind? Maybe there's some truth to that after all.
7777777phil•44m ago
Today I learned that Gemini can now natively embed video..

Cool Project, thanks for sharing!

kamranjon•25m ago
Does anyone know of an open weights models that can embed video? Would love to experiment locally with this.
sohamrj•6m ago
Not aware of any that do native video-to-vector embedding the way Gemini Embedding 2 does. There are CLIP-based models (like VideoCLIP) that embed frames individually, but they don't process temporal video. you'd need to average frame embeddings which loses a lot.

Would love to see open-weight models with this capability since it would eliminate the API cost and the privacy concern of uploading footage.

SpaceManNabs•18m ago
> No transcription, no frame captioning, no intermediate text.

If there is text on the video (like a caption or wtv), will the embedding capture that? Never thought about this before.

If the video has audio, does the embedding capture that too?

sohamrj•10m ago
Yes to both. The embedding is over raw video frames, so anything visible (text, signs, captions) gets captured in the vector. And Gemini Embedding 2 extracts the audio track and embeds it alongside the visual frames. So a query like 'someone yelling' would theoretically match on audio. My dashcam footage doesn't have audio though, so I haven't tested that side yet.
nullbyte•6m ago
What a brilliant idea! is this all done locally? That's incredible.

Show HN: Gemini can now natively embed video, so I built sub-second video search

https://github.com/ssrajadh/sentrysearch
65•sohamrj•2h ago•25 comments

Show HN: Email.md – Markdown to responsive, email-safe HTML

https://www.emailmd.dev/
24•dancablam•1h ago•2 comments

Show HN: TopMail – LaunchFest Winner, $20/Mo Unlimited Email, Coding Agents API

https://www.topmail.so
3•njarecki•1h ago•7 comments

Show HN: Cq – Stack Overflow for AI coding agents

https://blog.mozilla.ai/cq-stack-overflow-for-agents/
196•peteski22•1d ago•83 comments

Show HN: Jelly – SSH Social Hangout

2•jellyshelly•1h ago•0 comments

Show HN: Visualizing Apple Health workout data (stats, trends, insights)

https://apps.apple.com/us/app/streakout-workout-stats/id6758457318
2•toni88x•1h ago•0 comments

Show HN: Mapping the most convenient ways to meet someone with public transit

https://www.commutometer.com/meet-in-sanfrancisco
5•acavailhez•2h ago•0 comments

Show HN: Craton HSM – A memory-safe PKCS#11 software HSM in Rust

https://github.com/craton-co/craton-hsm-core
3•victor-craton•3h ago•3 comments

Show HN: Think Fast. Type Faster

https://wordsnap.up.railway.app/
4•oyahud•4h ago•1 comments

Show HN: MangoWave – Open-source browser audio visualizer

https://mangowave.app/
3•lmascari•4h ago•2 comments

Show HN: a Rust Redis GUI that doesn't freeze on 100k keys

https://github.com/vicanso/zedis
2•vicanso•5h ago•0 comments

Show HN: AI agent got 237 rules from another agent, still made the same mistakes

https://github.com/getcalx/oss
3•spenceships•5h ago•0 comments

Show HN: Off By – a daily game about how wrong we are about the American economy

https://offby.io
6•offby99•5h ago•3 comments

Show HN: The King Wen Permutation: [52, 10, 2]

https://gzw1987-bit.github.io/iching-math/
61•gezhengwen•1d ago•27 comments

Show HN: Running AI agents across environments needs a proper solution

https://github.com/liquidos-ai/Odyssey
7•human_hack3r•5h ago•5 comments

Show HN: zbot – an AI agent for MCU with Telegram chat (runs on Zephyr or Linux)

https://github.com/LingaoM/zbot
4•menggithub•5h ago•1 comments

Show HN: Danube – AI Tools Marketplace

https://danubeai.com
2•preston25•6h ago•1 comments

Show HN: Offline-first UK train planner

https://railraptor.com
3•marcusdev•7h ago•2 comments

Show HN: ProofShot – Give AI coding agents eyes to verify the UI they build

https://proofshot.argil.io/
67•jberthom•9h ago•52 comments

Show HN: Revise – An AI Editor for Documents

https://revise.io
81•artursapek•2d ago•72 comments

Show HN: Littlebird – Screenreading is the missing link in AI

https://littlebird.ai/
45•delu•1d ago•22 comments

Show HN: Kern – One agent. One folder. One mind. Every channel

https://github.com/oguzbilgic/kern-ai
4•obilgic•10h ago•1 comments

Show HN: Chat with an exhaustive geopolitical simulation of the 2026 Iran War

https://notebooklm.google.com/notebook/4cf9474f-194d-4607-8953-8ee84a9e66e0
4•hrishirc•3h ago•3 comments

Show HN: WhyThere.life – Compare cities side-by-side to decide where to move

https://whythere.life
15•daversa•20h ago•15 comments

Show HN: Atomic – Self-hosted, semantically-connected personal knowledge base

https://github.com/kenforthewin/atomic
147•kenforthewin•2d ago•24 comments

Show HN: Agent Kernel – Three Markdown files that make any AI agent stateful

https://github.com/oguzbilgic/agent-kernel
41•obilgic•1d ago•19 comments

Show HN: AgentDrive – Persistent file storage for AI agents

https://www.getagentdrive.com
3•itstomo•14h ago•0 comments

Show HN: We built a terminal-only Bluesky / AT Proto client written in Fortran

https://github.com/FormerLab/fortransky
145•FormerLabFred•3d ago•82 comments

Show HN: Sonar – A tiny CLI to see and kill whatever's running on localhost

https://github.com/RasKrebs/sonar
203•raskrebs•4d ago•80 comments

Show HN: Duplicate 3 layers in a 24B LLM, logical deduction .22→.76. No training

https://github.com/alainnothere/llm-circuit-finder
265•xlayn•5d ago•81 comments