Skip to content
Made with Jev
Submit

X · by Todd Dailey

A doorbell camera on Jev 1.14

Jev found all 23 people in the frames. The old pipeline found 14.

Todd Dailey on X

Todd Dailey

@twid

Tried @typesafeai’s jev 1.14 preview on my doorbell cam’s alert-to-Discord pipeline. My old setup: Gemma 4 26B describes the frame, then DeepSeek V4.1 Flash writes the caption. It caught only 14 of 23 people in recent motion events and missed the mail carrier. Asking jev “is a person in this frame?” across 4 frames of each event got all 23 in a test, with zero false alarms, in about 200 ms. (!) The new jev Box API then crops to the person, so Gemma can describe them far more reliably (21/23, up from 16). Concept to live in less than an hour. This replaces a LOT of AI calls and bounding box tools. The jev call costs about $0.00008 per event. (!) Things are changing so fast. 🚀🚀🚀

an illustration of the jev model catching a face. it says 23 of 23, a tiny fast judge catches every visitor at the door and boxes them in, and show a mail person handing mail to a robot.

Oct 15, 2026 · 16 likesOpen on X

Todd Dailey put the Jev 1.14 preview into his doorbell camera’s alert-to-Discord pipeline. His old setup had Gemma 4 26B describe the frame and DeepSeek V4.1 Flash write the caption, and it caught 14 of 23 people in recent motion events. Asking Jev “is a person in this frame?” across 4 frames of each event got all 23 in his test, with zero false alarms, in about 200 ms. The Box answer then crops to the person, and Gemma’s descriptions went from 16 of 23 to 21 of 23.

Open the source

More like this